Smart sensors for Indoor Environmental Quality in residential smart buildings: a review

International Journal of Sustainable Energy ISSN: 1478-6451 (Print) 1478-646X (Online) Journal homepage: www.tandfonline.com/journals/gsol20 Smart sensors for Indoor Environmental Quality in residential smart buildings: a review Andrea Alongi , Luca Pacileo , Mustafa Muthanna Najm Shahrabani , Paulius Spūdys , Rossano Scoccia & Livio Mazzarella To cite this article: Andrea Alongi , Luca Pacileo , Mustafa Muthanna Najm Shahrabani , Paulius Spūdys , Rossano Scoccia & Livio Mazzarella (2025) Smart sensors for Indoor Environmental Quality in residential smart buildings: a review, International Journal of Sustainable Energy, 44:1, 2578592, DOI: 10.1080/14786451.2025.2578592 To link to this article: https://doi.org/10.1080/14786451.2025.2578592 © 2025 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Published online: 20 Nov 2025. Submit your article to this journal Article views: 39 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=gsol20 INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 2025, VOL. 44, NO. 1, 2578592 https://doi.org/10.1080/14786451.2025.2578592 REVIEW ARTICLE Smart sensors for Indoor Environmental Quality in residential smart buildings: a review Andrea Alongi a , Luca Pacileo a , Mustafa Muthanna Najm Shahrabani b , Paulius Spūdys b , Rossano Scoccia a and Livio Mazzarella a a Politecnico di Milano, Department of Energy, Milano, Italy; b Kaunas University of Technology, Faculty of Civil Engineering and Architecture, Kaunas, Lithuania ABSTRACT The 2024 revision of the EU Energy Performance of Buildings Directive recognizes indoor environmental quality (IEQ) as a key complement to energy efficiency in promoting sustainable buildings and ensuring occupant comfort and well-being. This review approaches the subject of smart buildings from a multidisciplinary perspective, with a focus on residential applications. It examines four IEQ components: indoor air quality, thermal comfort, visual comfort and acoustic comfort. The discussion begins with technical standards and rating schemes, emphasising the physiological and psychological impacts of environmental conditions. It then investigates the state of research on smart sensors and IoT technologies, followed by recent advances in building management systems, particularly the integration of artificial intelligence for adaptive comfort control. Finally, the paper outlines future directions, including personalised comfort models, standardised assessment methods, scalable and interoperable sensor networks and privacy-preserving data strategies. ARTICLE HISTORY Received 1 August 2025 Accepted 14 October 2025 KEYWORDS Indoor environmental; quality comfort model; smart buildings; smart sensor; building management system; internet of things 1. Introduction In the past decades, most of the focus related to building design and construction has been on improving energy efficiency. However, the advent of sustainably certified buildings does not seem to be conclusively correlated with a comparable improvement in occupant satisfaction with indoor spaces (Asmar, Chokor, and Srour 2014; Geng et al. 2019; Pastore and Andersen 2019). Therefore, the notion of IEQ has recently become an additional pillar in the design and operation of modern buildings. It includes IAQ, TC, VC and AC and directly affects occupant health and well-being while having an indirect impact on energy performance and sustainability (Deng et al. 2024; Rupp, Vásquez, and Lamberts 2015). This paradigm is incorporated into the last revision of the EPBD (2024/1275) (European Commission 2024) published in May 2024, which highlights the importance of IEQ along with energy efficiency, introducing the for occupant-centric metrics, digitisation, and smart technologies in the built environment. This latter aspect is achieved by the introduction of the novel SRI, which is defined in the EPBD as a way to ‘measure the capacity of buildings to use information and communication technologies and electronic systems to adapt the operation of buildings to the needs of the occupants and the grid and to improve the energy efficiency and overall performance of buildings’ (European Commission 2024). Therefore, the definition of the SRI emphasises the importance of automation in buildings, which can be improved by integrating smart sensors with IoT networks and BMS. These devices can collect real-time environmental data and, along with the feedback of the occupants, support adaptive control algorithms, allowing buildings to adapt proactively and dynamically to variable conditions (Dong et al. 2019; Genkin and McArthur 2023). However, the deployment of smart sensor networks presents challenges, such as sensor accuracy, spatial variability of environmental parameters, privacy and interoperability. These © 2025 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4 .0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with their consent. CONTACT Rossano Scoccia [email protected] potential issues can be more prevalent in residential applications because of the greater variability in user behaviour and needs and architectural layouts (Calì et al. 2015; Clements et al. 2019). Given this context, this review the state of the research on smart sensors in relationship to IEQ monitoring in buildings, with a particular focus on residential applications. Section 3 provides definitions of IEQ and its main components, along with their relevance in terms of the health and well-being of occupants, the technical standards currently in force and the different models and schemes from the literature. Section 4 focussed on the concept of smart building, analysing the available definitions and the main features. Section 5 is dedicated to smart sensors and includes an overview of network configurations from the literature and an analysis of the measurement devices dedicated to all the variables required for smart building operation. Section 6 describes the development of BMS, including the adoption of AI and ML techniques. Finally, Section 7 concludes the paper with a discussion of the challenges and future developments that have emerged from the review of the literature, including the need for standardised assessment methods, improved sensor accuracy, and scalable, privacy-preserving solutions. Ultimately, this review highlights the multidisciplinary nature of the subject, linking technical innovations to the regulatory framework, highlighting the definition of IEQ and its components, their relevance in terms of human comfort and health, and the available rating schemes, and aims to provide support to researchers, technicians and policymakers in developing strategies and technologies for smart building construction, focussing on energy efficiency and occupant satisfaction and well-being. 2. Methods This study focusses on the use of smart sensors to assess and guarantee adequate IEQ conditions for occupants in smart buildings, with specific attention to residential buildings, and aims to highlight the state of the research, along with the challenges that must be addressed to advance the efficacy of these systems and facilitate their adoption. Therefore, it addresses multiple topics, from the notion of IEQ, discussing its importance and application, to the available sensing technologies and their integration into IoT networks and BMS. Research for this work has initially been conducted by querying both the Scopus and Web of Science databases, using the following keywords in various arrangements and combinations: IEQ, rating system, smart sensor, healthy building, case studies, digital twin, smart building, and BMS. The search results include original research papers, review papers and conference proceedings published in English from 2010 (with only a few exceptions) to 2024 (last update August 2024) and belong to the categories of engineering, environmental science, energy and computer science. After eliminating duplicates, the reference list has been expanded via a two-step process: first, through the snowball technique (a.k.a. citation search (Hirt et al. 2024)), and then via a more targeted search, to add information about specific subjects. The entire list was then screened for relevance to the topics studied in this review. Finally, European and US technical standards relevant to IEQ assessment and rating have been included, along with the data sheets of the devices presented in the literature, to provide the reader with full access to the list of their features. The whole process is summarised by the flow diagram represented in Figure 1. Keyword relevance has been investigated considering the final selection of references from the scientific literature by highlighting co-occurrences. The outcome of this analysis is a network map of all keywords with at least five occurrences produced using the VOSviewer tool (van Eck and Waltman 2010), as shown in Figure 2. The size of each label represents its weight, while the distance between nodes is related to the strength of their connection, and the thickness of the links indicates the probability of co-occurrence. Finally, the nodes are coloured according to the average normalised citation number (i.e. the number of citations of documents related to a keyword divided by the average number of citations of the documents published in the same year to account for potentially higher citation numbers for older references). Figure 2 shows that indoor thermal comfort and indoor air quality are the most prevalent subjects among those related to indoor environmental quality, with a number of citations above average, demonstrating the research interest in those topics. At the same time, the reference list includes a significant number of manuscripts dedicated to artificial intelligence, occupancy and building management system, which have received an average number of citations, considering the normalization. 2 A. ALONGI ET AL. Finally, the research papers and the standards among the final reference list can be grouped according to their pertinence for the main topics investigated in this manuscript (IEQ, standards and schemes, smart buildings, smart sensors and IoT, BMS and case studies) to analyse their distribution in time: Figure 3 shows that most of them have been published after 2015, with a growth in interest over time, and that the scientific production dedicated to smart buildings and the description of case studies is even more recent overall. In general, Figure 3 shows that most of the references considered in this review have been published within the last ten years. 3. A comprehensive look at indoor environmental quality This section is dedicated to the general concept of IEQ: starting from its latest definition and introducing the main components, an overview of the consequences related to indoor environmental conditions is presented, highlighting the comfort, health and productivity of the occupants. Finally, the standards and assessment schemes for IEQ are presented. 3.1. Definition The IEQ refers to the indoor conditions in a building and is related to both the comfort and the health of the occupants. Its main components are TC, IAQ, VC and AC (Mui et al. 2016). More precisely, this concept describes an integral state of an occupant's subjective response to indoor environmental parameters such as the air temperature, humidity, CO2 concentration, horizontal illumination, sound pressure Records identified from Scopus and WoS: Original Research and Reviews (n = 439) Additional search Original Research and Reviews (n = 67) National and International Standards (n = 25) Datasheets (n = 17) References included in the paper Total* (n = 250) - Datasheets (n = 17) - National and International Standards (n = 25) - Original Research and Reviews (n = 208) Records collected through Snowball Technique Original Research and Reviews (n = 114) Records removed before screening Duplicate records removed (n = 30) Records addressed in the first screening Total (n = 632) - Datasheets (n = 17) - National and International Standards (n = 25) - Original Research and Reviews (n = 590) Records removed after first screening Original Research and Reviews removed (n = 342) Records addressed in the second screening Total (n = 290) - Datasheets (n = 17) - National and International Standards (n = 25) - Original Research and Reviews (n = 248) Records removed after second screening Original Research and Reviews removed (n = 40) IDENTIFICATION SCREENING INCLUDED Figure 1. Flowchart describing the process followed to collect references for this review, divided into identification, screening and included. *Note: The total number of references excludes two works related to the search methodology followed in this work. INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 3 Figure 2. Keyword co-occurrence network, coloured according to the average normalised number of citations. ≤2010 2011-2012 2013-2014 2015-2016 2017-2018 2019-2020 2021-2022 2023-2024 year of publicatiton 0 5 10 15 20 25 number of records Indoor Environmental Quality Standards and Schemes Smart Buildings Smart Sensors and IoT networks Building Management Systems Case Studies Figure 3. Number of research papers and standards included in the reference list, divided by year of publication and pertinence for the main topics investigated in this work. Each reference can potentially be considered relevant for more than one topic. 4 A. ALONGI ET AL. level and local air velocity (Mendell 2003). Moreover, this notion can also be extended to include daylight and views, pleasant acoustic conditions and occupant control over lighting and thermal settings. These elements contribute collectively to the whole perception (Deng et al. 2024) and can influence people's level of comfort (Sim et al. 2016). Additional aspects, such as layout, available space, furniture, and interior design, can also affect perceived comfort and well-being, although they are generally excluded from standardised comfort assessments (Zhang, Mui, and Wong 2023). According to Standard ANSI/ASHRAE 55-2004 (2004), TC is defined as, that condition of mind which expresses satisfaction with the thermal environment. According to Ganesh et al. (2021) and Navada, Adiga, and Kini (2013), the parameters related to TC can be divided as follows: a. Environmental factors, including the indoor air temperature, relative humidity, air velocity and mean radiant temperature; b. Personal factors that depend on the individual characteristics of a person, including metabolic rate, clothing insulation, sex, age, and weight. In addition, climatic, geographic and cultural factors should also be considered in the evaluation of TC but are currently overlooked in most assessment methods (Horr et al. 2016). As far as IAQ is concerned, ASHRAE Standard 62.1-2016 (2016) defines good air quality as follows: air in which there are no known contaminants at harmful concentrations as determined by cognizant authorities and with which a substantial majority (80% or more) of the people exposed do not express dissatisfaction. In fact, IAQ is a multidisciplinary phenomenon and is determined by the many pathways in which chemical, biological and physical contaminants eventually become a part of the total indoor environmental composition (Tham 2016). Dealing now with VC, the European standard EN 12665 defines it as: a subjective condition of visual well-being induced by the visual environment (En 12665: 2024). According to Carlucci et al. (2015), its assessment includes the evaluation of: a. The amount of light; b. The uniformity of light; c. The quality of light in rendering colors; d. The prediction of the risk of glare for the occupants. The study conducted by Ko et al. (2022) revealed that VC can also be improved by window view quality. Its assessment should consider daylight, views and building orientation but should also be flexible enough to not prevail over other requirements. Finally, AC is defined in Frontczak and Wargocki (2011) as: a state of contentment with acoustic conditions. The main factors include the sound pressure level, exposure time, frequency composition and repetition over time (Rocca et al. 2022). However, overall perception can also be influenced by the urban context, envelope acoustic performance, person-related factors, situational factors and the environmental context (Torresin et al. 2019). 3.2. Importance People spend 90% of their time indoors, making IEQ a central aspect of modern building design. In this context, the concept of a people-centred building has been promoted in the literature (Jin et al. 2018), considering the health, comfort, well-being and productivity of the occupants as effects of improved comfort levels (Devitofrancesco et al. 2019). According to Asadi, Mahyuddin, and Shafigh (2017), poor IEQ is caused mainly by elevated IAP concentration, indoor temperature and relative humidity. These conditions can cause physiological stress, reduced cognitive function and chronic health problems (Heidary et al. 2023), as well as an increased possibility of developing SBS (Zhang, Mui, and Wong 2023). Among the components of IEQ, indoor air INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 5 quality (IAQ) plays a predominant role in the protection of human health, since it directly influences shortand long-term physiological results. Indeed, inadequate IAQ levels can lead to various consequences: some pollutants are considered more hazardous and are associated with respiratory diseases, cancer, immune deficiencies, and organ diseases affecting the liver and kidneys (Tham 2016). In particular, the study conducted by Patino and Siegel (2018) highlights the elevated concentration of PM2.5 in social houses due to cigarette smoking in buildings, revealing it as the primary cause of health issues related to IAQ. Furthermore, poor IAQ can facilitate the transmission of infectious diseases and increase the incidence of health symptoms in occupants, such as mucous membrane irritation, airway issues, headaches, a lack of concentration, fatigue, allergic reactions, asthma, etc. (Altomonte et al. 2020). Other pollutants may not be directly harmful but still negatively influence comfort, productivity and overall performance (Xiang et al. 2013). For example, according to Wolkoff (2013), odour appears to be one of the IAQ factors capable of influencing occupant perception, despite not necessarily being related to adverse health effects. Finally, it is important to clarify that the research shows that symptoms are correlated with the level and duration of exposure (Snyder et al. 2013). Similarly, uncomfortable thermal conditions can cause illnesses or other effects, such as shivering and sweating, which are more closely related to the well-being of the occupants than to severe health issues (Sim et al. 2016). The work of Choi and Zhu (2015) indicates that VC is a key component of IEQ, particularly in workplace environments where occupant health and productivity are critical. Other studies have revealed that glare and reflected light negatively impact user satisfaction in computer-based tasks (Choi, Loftness, and Aziz 2012), while natural light provides physical and psychological benefits to building occupants (Dahlan and Eissa 2015). Moreover, our eyes are not only responsible for vision but also crucial in stimulating our body's circadian system, which sets the pace of almost every process in our bodies (Altomonte et al. 2020). According to Sakellaris et al. 2016, the results obtained through a cross-European survey demonstrate that AC is the most important IEQ component associated with occupant comfort, particularly satisfaction with overall noise. Noise levels represent the most significant factor that affects sleep quality, concentration, and overall performance (Muzet 2007). Hongisto (2005) explored different sources of disturbance: a predictive model demonstrated that it is not the sound level of speech that determines its distracting power but its intelligibility. On this matter, the STI quantifies how clearly a spoken message can be perceived and understood in an environment affected by background noise and reverberation, and the results show that high STI values correspond to a reduction in the performance of the occupant. Finally, a review by Rupp, Vásquez, and Lamberts (2015) highlights that the IEQ strongly influences human performance and productivity: maintaining high comfort conditions improves the quality of life and productivity of work. Indeed, the work by Fisk (2000) estimates the annual economic benefits of improved IEQ in the United States, including reduced health-related costs and increased worker performance, thereby underscoring the significant potential of this research domain. Therefore, similar to energy efficiency improvements, a significant push toward better IEQ could be driven by financial incentives Parkinson, and de Dear (2019), highlighting the costs associated with inadequate indoor conditions, particularly in office environments (Zhang et al. 2023). The challenge lies in achieving high IEQ standards according to occupant perceptions without increasing energy consumption (Choi and Yeom 2017; Kaushik et al. 2020). A potential solution is proposed in the manuscript by Deng et al. (2024), where the authors suggest combining smart sensors with occupant feedback as an effective approach to achieve both energy efficiency and indoor comfort. This topic will be discussed in more detail in Section 4. Another suggestion proposed by Qabbal, Younsi, and Naji (2022) recommends to incorporate IEQ during the design phase, especially for AC and VC, investigating envelope geometry, technology and indoor space design (Horr et al. 2016). Table 1 summarises the relationship between IEQ category effects on occupants. 3.3. Standard and schemes The IEQ is linked to a wide range of parameters, constraints and assessment methods. The overall level of IEQ in buildings depends on the interactions among various parameters (Nimlyat 2018). The work by Wei 6 A. ALONGI ET AL. et al. (2020) provides a review of fourteen green building certification schemes to assess the IEQ levels in offices and hotels. The outcome counts 19 parameters to assess TC, 39 for IAQ, 20 for AC and 12 for VC. Given this complexity, the amount of data required for a comprehensive assessment of IEQ often exceeds the capabilities of currently available sensors in terms of economic and technological feasibility. A detailed discussion is provided in Section 5. To simplify the process, only a limited number of parameters are typically monitored, according to their importance and influence on IEQ. For instance, (Jin et al. 2018) proposed temperature, humidity, illuminance, CO 2 , VOC and PM as the main parameters. However, other studies have highlighted the limitations of relying on a narrow set of variables for IAQ assessment; For instance, (Pastore and Andersen 2019) stated that relative humidity and CO 2 concentration are insufficient to accurately reflect occupant satisfaction. Another major challenge in this field is the development of assessment frameworks that are able to incorporate occupant feedback. To address this, (Geng et al. 2019) propose to separate IEQ assessment into an objective part, which compares measurements to thresholds provided by standards, and a subjective part aimed at evaluating occupants' perceptions registered through surveys. The work by Meir et al. (2009) highlights the importance of POE as a tool to guarantee that new buildings are able to respect increasingly demanding standards of comfort, safety, cost-effectiveness and sustainability. Indeed, in (Meir et al. 2020), POE is used to identify potential flaws in various phases of the building life (i.e. design, construction, commissioning and building/user interaction) while indicating useful corrective feedback. These questionnaires are based on human perception (Ganesh et al. 2021), and their objective is to assess psychological factors. Although individual variability may limit their comprehensiveness, well-designed surveys remain highly effective. The BUS occupant survey Bordass, Leaman, and Eley (2016), For instance, systematically translates occupant feedback into actionable data, enabling case-specific analyses, benchmarking across a broader database, and providing robust support for decisionmaking. In this sense, user feedback constitutes an integral part of a structured process that improves both individual buildings and collective knowledge. Subjective evaluation techniques are also included in Annex F of CEN/TR 16798-2 (CEN 2019). General IEQ assessment based on its four components relies on EN 16798-1:2019 (Standard CEN/TR 16798-2:2019). The standard classifies dwellings into four different categories based on the occupants PPD. In parallel, ANSI/ASHRAE/IES Standard 90.1-2016 (2016) is considered equivalent to EN 16798-1:2019 and ensures compliance with energy requirements. However, as highlighted in (Wargocki et al. 2021), a clear reference framework is needed to define which variables should be measured and how different Table 1. IEQ categories and effects on occupants. References IEQ category Environmental parameters Health effects Well-being and productivity (Sim et al. 2016) TC Air temperature, relative humidity, air velocity and mean radiant temperature Illnesses Shivering and sweating (Tham 2016) IAQ Indoor air pollution (IAP) Respiratory diseases, cancer, immune deficiencies, and organ diseases affecting the liver and kidneys Influence comfort, productivity, and overall performance (Patino and Siegel 2018) IAQ PM 2.5 Biggest chronic health impact among common residential indoor pollutants (Choi and Zhu 2015) VC Glare and reflected light Reduce user satisfaction (Muzet 2007) AC Noise reduce sleep quality, concentration, and overall performance (Hongisto 2005) IEQ Speech transmission index (STI) Reduction in occupant’s performance (Heidary et al. 2023) IEQ Elevate indoor air pollution (IAP) concentration, indoor temperature, and relative humidity Chronic health issues Physiological stress, reduced cognitive function (Rupp, Vásquez, and Lamberts 2015) IEQ Enhances quality of life and work effectiveness (Zhang, Mui, and Wong 2023) IEQ Sick building syndrome (SBS) INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 7 outcomes should be weighted. Since EN 16798-1:2019 builds upon the reference standards of each individual domain, namely, TC, IAQ and VC, a more detailed overview of these component-specific standards is provided. ISO 7730 (ISO 2005) and ASHRAE 55 (Standard ANSI/ASHRAE 55-2004 2004) are the reference standards for the evaluation of TC and are widely adopted in scientific research (Daum, Haldi, and Morel 2011; Navada et al. 2013; Nicol and Humphreys 2002; Rupp, Vásquez, and Lamberts 2015; Khovalyg et al. 2020). These standards define acceptable thermal environments based on the PMV, which represents the average value predicted by the subjective judgement of a group of people in a given environment, and PPD, which refers to a quantitative measure of thermal comfort. With respect to TCs, adaptive criteria are aimed at setting acceptable indoor temperatures in buildings without mechanical cooling. The method can be applied under specific conditions based on the building type, occupant behaviour, metabolic rate, clothing insulation and seasonal period. The purpose is to have a dynamic indoor temperature setpoint that permits energy savings and enhances occupant involvement and adaptation. The assessment is conducted through the evaluation of the operative temperature, as prescribed by CEN/TR 16798-2 (CEN 2019). However, several studies have pointed out limitations in their accuracy, particularly due to the variability of human perception influenced by different cultural habits, climate conditions and individual factors (Niza and Broday 2022). Among these individual factors, clothing insulation plays a crucial role in perceived TC. In this context, Zhang (2010) explored the impact of clothing on thermal sensation by applying a modified Gagge model to simulate transient heat and moisture transport through clothing systems, offering a more dynamic and personalized approach to TC evaluation Ghaddar, Ghali, and Chehaitly (2011). Related to the variability of human perception, (Kim, Schiavon, and Brager 2018a; Kim et al. 2018b) introduced the concept of personal comfort model for TC with the aim of predicting an individual's thermal comfort response. This approach identifies the specific comfort needs of each individual and provides the desired environmental conditions required to satisfy them. This information can then be used to operate the HVAC system, as indicated by Jung and Jazizadeh in their work (Jung and Jazizadeh 2019). The main features of a personal comfort model are as follows: a. Treating the individual as the unit of analysis; b. Collecting both direct feedback and personal data from occupants; c. Measuring environmental parameters; d. Prioritizing cost-effective and easily obtainable data; e. Adopted a data-driven approach in order to increase the flexibility and adaptability of the model. Results showed improvements in the prediction accuracy (20%–40% higher than that of convectional comfort models) and enhanced data diversity. Issues arise when individual feedbacks are not enough to train the model, mainly because of the decrease in occupant participation. Another work (Choi and Yeom 2017) focused on personal models reveals the possibility to assess TC through skin temperature measurements on the wrist (front), upper arm, and waist. Gender and BMI, which refer to the ratio between body weight and height of the human subject, were also considered. The study carried out in an experimental chamber reveals a level of accuracy of 95% compared to the responses registered with surveys. However, issues arise due to the discomfort and invasiveness of the body sensors. IAQ assessments refer to ANSI/ASHRAE Standard 62.1-2022 (Standard ANSI/ASHRAE Standard 62.12016) and ANSI/ASHRAE Standard 62.2-2022 (Standard ANSI/ASHRAE 62.22022), the latter being specifically focused on residential buildings (Calì et al. 2015; Berger et al. 2022; Khovalyg et al. 2020), by calculating the minimum requirements for ventilation. Additionally, EN 16798-1 (Standard CEN/TR 16798-2:2019 2019) specifies ventilation rates with three different methods: • Perceived IAQ: based on comfort criteria. The total ventilation rate is calculated by combining two components: the ventilation required for occupant bio effluents and the ventilation required for building material/system emissions; • Substance concentration limits: these are based on health criteria. The required ventilation rate to dilute an individual substance is calculated using a steady-state mass balance formula based on the pollutant's generation rate and the maximum permissible guideline concentration; • Predefined ventilation flow rates: minimum rates estimated to meet both health and perceived air quality, typically expressed as flow rates per person, per unit floor area, or air change rates; 8 A. ALONGI ET AL. and peak energy consumption forecasting while also supporting data-driven control algorithms for occupancy prediction and HVAC unit status inference (Shapi et al. 2021). Advanced deep learning approaches, including hybrid models combining CNN and RNN, are being integrated into BMS platforms for short-term heat energy consumption prediction, supporting more sophisticated energy management strategies in smart buildings (Sharma et al. 2024). Traditionally, smart buildings have made use of various sensors to gather information on occupancy patterns and optimize energy-consuming facilities. However, despite the advancements in BMS technology, several challenges and limitations persist. First, sensor accuracy and calibration are two major challenges that can deeply affect the quality and reliability of collected data (Painter, Brown, and Cook 2012; Akkaya et al. 2015). For instance, miscalibrated sensors might prompt the HVAC or lighting systems to change status when they should not, compromising energy efficiency and occupant comfort (Painter, Brown, and Cook 2012). Furthermore, privacy concerns exist, especially regarding the use of occupancy detection systems with cameras or other potentially invasive technologies (Shih 2014; Akkaya et al. 2015). Finally, the deployment of effective smart building solutions is further complicated by integration challenges, such as the need for BMS compatibility, since manufacturers do not necessarily use the same standards (Habiba et al. 2024). Moreover, there is an ongoing discussion around the extent that green building certifications, including IEQ, ensure real comfort conditions. Although studies have focused on the certification of green buildings, it has been observed that it does not necessarily guarantee comfortable conditions in all aspects, since TC and AC are not considered consistently (Horr et al. 2016; Karimi et al. 2023). The literature reveals diverse viewpoints on whether BMS achieves the goals of energy efficiency and comfortable indoor environments. Advanced BMS functionalities, including adaptive systems control and real-time data analytics, are associated with remarkable results on both fronts (Li et al. 2019; Chen et al. 2021). For example, datadriven adaptive HVAC control results in decreased thermal dissatisfaction in the range of 5%−40%, together with energy savings of 15%−33% (Papadopoulos et al. 2023). Nevertheless, other works highlight several limitations, such as the dependence on correct sensor data, as well as the risk of privacy breaches with invasive surveillance tools (Painter, Brown, and Cook 2012; Akkaya et al. 2015). Various technological advancements in digital twins and AI-driven predictive models, among others, are providing solutions to these challenges, since a digital twin can provide near real-time data and insights to facilitate maintenance and optimization of building operations (Yang et al. 2022; Puiu and Fortis 2024). The evolution of BMS is moving towards highly adaptive, occupant-centric systems based on emerging technologies (e.g. AI, IoT, digital twins) to enhance its features and capabilities. Adaptive control strategies will further gain traction, dynamically adjusting building operations with real-time data and feedback from occupants to find the best trade-offs between energy efficiency and comfort (Salamone et al. 2018; Papadopoulos et al. 2023). 5. Smart sensors The effectiveness of BMS highlights the importance of accurate and reliable data collection, along with properly designed IoT networks. This section is dedicated to the sensing technologies available for IEQ monitoring in smart buildings. After a general overview, the key sensors dedicated to the main environmental variables (i.e. air quality, temperature and relative humidity, acoustics, and light) are investigated. Although occupancy evaluation is not strictly involved in IEQ monitoring, it is included in this literature analysis because of its relevance in energy savings and personalized comfort (particularly in residential settings). Finally, all-in-one and wearable smart sensors have been investigated due to the recent growth in importance on the market. 5.1. Overview Sensors are at the core of a smart building: they collect data on environmental parameters and occupation, providing valuable information to assess IEQ and guarantee comfort conditions while optimising system INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 15 management to reduce energy consumption (Dong et al. 2019). In detail, the notion of smart sensors was introduced in the literature around the early 2000s: they are nodes in an IoT network, and their hardware may include memory, a computational device and a wireless communication device, along with the sensing element itself. They can also include software for monitoring and signal processing and can provide control signals to actuators connected to the network (Lim 2001). Therefore, smart sensors can also have the ability to post-process the output of the sensing element and perform self-diagnosis (Bayart 2001) and self-calibration (Kenny 2005) procedures. As stated in (Mozaffari et al. 2019), smart sensors constitute the so-called edge layer of the overall network: due to their computational capabilities, they can process data and limit the bandwidth required for transfer to other layers. The same work lists two other requirements to allow the integration of smart sensors into an IoT network: a. They need to be logically connected to each other to combine data and improve precision; b. They need to be highly energy efficient. Since the effectiveness of an IoT network relies on its architecture, which needs to be scalable and able to manage, transfer and analyse collected data, much focus has been dedicated to this topic in recent years. An interesting approach applied to public buildings with an existing communication infrastructure was presented by Calvo et al. (2022): it is based on the highly flexible and scalable edge-fog-cloud paradigm, namely, sensing nodes, IEQ concentrators (local storage, configuration, etc.) and cloud services (e.g. data analytics, storage, prediction, weather forecasting, etc.). Similarly, the work by Kumar et al. (2021) suggests a general architecture layout that can be applied to any context and follows the evolution of the smart building over the years and is based on three layers: the sensing or perception layer, which is responsible for collecting data, including those related to occupant preferences (e.g. set points for temperature and relative humidity); the processing or network layer, which organises and processes information; the reproduction or application layer, which uses processed data to understand the interaction between users and equipment and, finally, improves the performance of the latter in delivering services to the users themselves. This framework aims to track the environmental quality and use of indoor spaces and improve building management and maintenance. A similar three-phase concept was presented by Li et al. (2019) and can be summarised as monitoring, diagnostic and intervention. The monitoring phase is mainly based on the real-time sensor network but can also include the collection of information on occupant satisfaction. The diagnostic phase is based on data-mining techniques and statistical analysis and is focused on identifying performance losses and providing possible interventions to improve energy efficiency, IEQ and occupant satisfaction. Finally, the intervention phase focusses on developing and validating optimisation strategies, such as automatic controls and occupant behaviour. Finally, Genkin and McArthur (2023) ascribe the smart sensor network to the real-time data zone of the building knowledge repository inside the B-SMART architecture. This zone needs to support high rates of data inserts and updates, along with fast operations when needed, and to store at least 24 h of collected information to allow for real-time analysis. For older data and technical specifications, the historical data zone is available inside the architecture. This approach treats smart buildings as autonomic cybernetic systems, which are computing environments capable of minimising the human intervention required in response to changes, either internal or external. In terms of overall efficacy in guaranteeing high levels of IEQ in smart buildings, a dedicated sensor network should be designed that considers the spatial distribution of environmental variables (Calvo et al. 2022). This issue is addressed in an experimental study by Clements et al. (2019), which evaluated the efficacy of a living lab research facility to conduct human-subject research: when thermal conditions are considered, the authors measured up to 3 °C of temperature deviation between the set-point and desk temperature when windows were not shaded and significant variation between desks due to layout, human presence, orientation, etc. The same work highlights spatial variability in lighting and auditory conditions variation in background sound levels between 32 and 43 dBA). The distribution of indoor pollutant concentrations can also be spatially non-uniform, as stated in Xiang et al. (2013). Such variability requires a sufficiently refined sensor network to allow adequate monitoring of environmental conditions, at least in the most commonly used locations inside an indoor space (Dong et al. 2019). This is also demonstrated in Mohammadi, Assaf, and Assaad (2024), where a set of four sensing stations is used to evaluate the spatial distribution of temperature in a compact space (3.2 × 5.2 m room) in real time. A more targeted approach can be adopted in working environments such as office buildings, where sensors can be located on desks to 16 A. ALONGI ET AL. evaluate environmental variables locally (Salamone et al. 2017; Parkinson et al. 2019). When IAQ is considered to evaluate personal exposure to pollutants, it presents the same kind of challenges due to the spatial and temporal variability of their concentration, as highlighted in Qabbal, Younsi, and Naji (2022). As far as lighting quality and visual comfort in office environment are concerned, the paper by Pandharipande and Caicedo (2015) presented a smart lighting system that involves luminaires equipped with occupancy and light sensors. Due to its granular application, it is able to mitigate energy consumption by actuating each luminaire separately only when and where required by the presence of a user and dimming the level to take into account the available daylight (to guarantee the minimum illuminance of 500 lux in occupied areas and 300 lux in unoccupied areas, as indicated by the standard EN 12464-1 (EN1 2021). However, despite the observed energy savings, the authors identify the role of daylight on lighting control as one of the technical challenges that would be mitigated by the introduction of automatically controlled blinds to lower glare. Finally, in recent years, smart sensors dedicated to residential settings have been made available on the market (LSI LASTEM S.R.L.- Sphensor TM; RENSON - Sense; leapcraft - AmbiNode; Fybra S.R.L. - Fybra Home): they can be located inside rooms according to the needs of the occupants and provide useful information about the IEQ. However, their placement is limited by the interior arrangement of the rooms, and readings might not reflect the conditions at the exact location of the occupants (for instance, sensor readings taken from one side of a room may not accurately reflect the environmental conditions on the other side). Building on the concepts of smart sensors, IoT layers and scalable architectures discussed above, the general control workflow of a smart building is illustrated in Figure 4. The diagram presents how heterogeneous sensor data (e.g. air quality, temperature, humidity, light, acoustics, and occupancy) are collected and transferred across edge-fog-cloud layers for further processing, storage and analysis. Furthermore, the analysis and control layer, which can constitute AIand BMS-driven control logic, enables optimized HVAC strategies, which are implemented via actuators (fans, pumps, valves, motors, etc.). Additionally, the workflow emphasizes the continuous feedback loop between occupants, sensing infrastructure, and intelligent control systems, ensuring optimal occupant comfort and well-being. 5.2. Key sensors The analysis of works in the literature dedicated to IEQ and smart buildings highlights the state-of-the-art and future trends in sensing devices that can be integrated into smart sensors, together with the sensors already available on the market. The purpose is to list the main technologies available to address several components of the IEQ, along with the potential limitations and issues that should be considered when effectively deploying IoT networks and smart sensors in residential buildings. Moreover, greater emphasis is given to devices that involve low-cost sensors oriented toward a do-it-yourself approach (Table 4), since the vast number of variables required for IEQ monitoring can lead to a growth in the complexity and cost of smart sensor networks, potentially hindering their diffusion in the market. Air Quality Temperature and Humidity Light Acoustics Occupancy All-in-one Wearable Sensors Data transfer/storage Analysis/Control BMS AI and Machine Learning HVAC control strategies Thermostat Actuators Fan Pump Valve Motor ... Occupant app Edge Fog Cloud Feedback Figure 4. IEQ control workflow. INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 17 5.2.1. Air quality IAQ in buildings has a direct effect on the health and well-being of their occupants. It is affected by several parameters, such as the concentrations of various gases (e.g. CO 2 , CO, NO x , SO x , etc.), VOCs and PM (Heidary et al. 2023). Therefore, different sensors are dedicated to each category, and a practical implementation of Smart Sensors in residential Smart Buildings would require a careful selection of the most relevant quantities to be measured. CO 2 is one of the key indicators related to IAQ (Wargocki et al. 2021): it can be used to address the efficacy of the ventilation system in providing fresh air intake and controlling the air change rate through a demand-controlled strategy (Nassif 2012) and is related to the outdoor concentration and occupant presence and activity as a consequence of their metabolism (Dong et al. 2019). For instance, the study by Laverge et al. (2015) shows that the highest concentrations observed in residential buildings are detected in bedrooms. Therefore, CO 2 concentration can be used as a control variable to manage the ventilation system since it can be used as a proxy for the occupancy level. In this way, it is possible to achieve a healthy indoor environment and mitigate energy consumption at the same time (Nassif 2012; Wang et al. 2018; Zikos et al. 2016; Wolf et al. 2019). However, when it is used for this purpose, the location of CO sensor installation becomes paramount (Zikos et al. 2016; Wolf et al. 2019), as discussed in § Section 5.2.5. Moreover, according to (Erickson and Cerpa 2010; Zikos et al. 2016), the slow response of CO2 sensors to build-up can make the ventilation control less effective, making their calibration and accuracy critical (Nassif 2012). Even though in past works electrochemical sensors were considered among Table 4. List of low-cost sensing elements for IEQ, including the works from literature where they are discussed (column Refs.). Measurement range and accuracy are reported, when available. Sensor Features References Notes Air quality K-30 (Datasheet K30 2022) NDIR CO 2 sensor 0…5000 ppm ± 30 ppm (Salamone et al. 2017; Pandharipande and Caicedo 2015) (Weyers et al. 2017) (Weekly et al. 2015) Equipped with automatic baseline correction Telaire T67xx modules ( Amphenol Sensors) NDIR CO 2 sensor 0…2000 ppm ± 30 ppm (Ekwevugbe et al. 2017) (Candanedo and Feldheim 2016) (Dong and Lam 2011) Equipped with self-calibrating algorithm S8 (Datasheet, S8 2022) NDIR CO 2 sensor 400…2000 ppm ± 70 ppm (Wolf et al. 2019) Maintenance-free, available in a low-power version CSS811 (Datasheet CCS 811 2020) MOX sensing element, directly measures TVOC concentration and evaluates eCO 2 0…1187 ppb TVOC 400… 8192 ppm eCO 2 (Calvo et al. 2022) Low-cost, calibrated to a typical mixture for indoor environment, baseline correction every 24 h, affected by low operational precision TGS2602 (Datasheet, TGS2602, 2018) MOX sensing element, VOCs 1…30 ppm (Snyder et al. 2013) (Wolf et al. 2019) This technology can be used to monitor CO, O 3 and NO x GP2Y1010AU0F (Datasheet, GP2Y1010AU0F, 2018) PM monitoring through light scattering 100 g m 3 ± 30% (Snyder et al. 2013) (Qabbal et al. 2022) It does not discern between PM 2.5 and PM 10 (GP2Y1030/31AU0F can detect both, DN7C3CA007/ DN7C3D015 is dedicated to PM 2.5 ) Temperature and Relative Humidity DHT22 (Datasheet, DHT22 2012) Capacitive temperature and relative humidity sensor 0…100% ± 5% RH −4080 °C ± 0.2 ˚°C T (Mui et al. 2016) (Qabbal et al. 2022) (Salamone et al. 2017) (Candanedo and Feldheim 2016) Reported accuracy do not agree between different works (values reported here are referred to data provided by the manifacturer) Telaire T9602 (Datasheet, Telaire T9602 2018) Capacitive polymer RH sensor, proportional to absolute T sensor 20…80% ± 2% RH 2040 C ± 0.5 ˚°C T (Weyers et al. 2017) Reliable low-cost option BME680 (Datasheet, BME680 2024) RH, T, VOC and pressure sensor 0…100% ± 3% RH 4085 °C ± 1°C T (Mohammadi et al. 2024) Compact and low-powerconsumption 4-in-1 sensor Light TSL2561 (Datasheet, TSL2561 2025) Photodiode 0.1…40000 lux (Mui et al. 2016) (Candanedo and Feldheim 2016) Separate detection of infra-red and full-spectrum light, comparable performance to laboratory grade instruments Acoustics Grove Sound Sensor (Datasheet, Grove 2015) Electret microphone, LM358 amplifier 40…90 dBA (Mui et al. 2016) (Parkinson et al. 2019) (Qabbal et al. 2022) (Zikos et al. 2016) 18 A. ALONGI ET AL. the most promising in terms of cost and durability (Park et al. 2003), the devices mentioned in more recent works are generally based on the NDIR operating principle, which is implemented by several commercially available sensing modules, such as the K-30 (Datasheet, K-30 2022) (used in Pandharipande and Caicedo 2015; Weyers et al. 2017; Weekly et al. 2015; Salamone et al. 2017) and various versions of the Telaire (Amphenol Sensors) sensor (mentioned in (Ekwevugbe et al. 2017; Candanedo and Feldheim 2016; Dong and Lam 2011). They feature measurement range equal to 0…5000 ppm and 0…2000 ppm, respectively, with an accuracy around ± 30 ppm, and both are equipped with a baseline correction or a self-calibration algorithm. NDIR is also implemented in the S8 (Datasheet, S8 2022) miniature CO 2 sensor (range 400… 2000 ppm, accuracy ± 70 ppm, maintenance-free according to the manufacturer) mentioned in the work by Wolf et al. (2019), which is available in a low-power version. Sensors based on this principle are generally compact and stable to fluctuations in temperature and relative humidity, while their sensitivity is affected by the path length, and the calibration procedure can pose some issues (i.e. self-calibrating single beam devices can assume 400 ppm as background CO 2 as default) (Snyder et al. 2013). Finally, another approach is the indirect measurement of the so-called equivalent CO 2 (eCO 2 ) performed by the CCS811 sensor (Datasheet, CCS811 2020): this quantity is tied to TVOCs concentration (which is measured directly) and is considered to provide an acceptable representation of the actual CO 2 concentration evolution. This sensor is used in (Calvo et al. 2022) due to its lower cost, but the authors have observed a lack of operational precision and plan to search for an alternative. Even though the CO 2 concentration is sometimes used as the sole indicator for IAQ (Devitofrancesco et al. 2019) and to quantify the need for fresh air in indoor spaces (Nassif 2012), there are other factors that need to be taken into account, especially in urban areas, where a higher ventilation rate can increase the indoor pollutant concentration (Kumar et al. 2016), hence the requirement for effective filtration. Moreover, only inspecting for CO 2 does not consider the emissions from building materials (Painter et al. 2012). This complex relationship, demonstrated by the measurements presented in (Qabbal, Younsi, and Naji 2022), highlights the necessity for more in-depth monitoring, including of pollutant gases, VOCs and PM. Indeed, beside CO2 , there are several potential variables to monitor when assessing IAQ: as an example, the TAIL protocol (Wargocki et al. 2021), aimed at evaluating the IEQ in offices and hotels, also includes the concentration of formaldehyde, benzene and PM 2.5 . This list was expanded further by Qabbal et al. in their experimental evaluation of a university building (Qabbal, Younsi, and Naji 2022) by considering VOCs and CO 2 concentrations as well. Finally, some works only analyse VOCs (Xiang et al. 2013; Calvo et al. 2022) or PM (Weyers et al. 2017). This lack of consistency suggests that there is not a general definition for a minimum set of variables to be continuously measured when addressing the IAQ of an enclosed environment, but it can depend on several factors (e.g. use, location, and construction technology, etc.). As far as residential buildings are concerned, an interesting review presented by Rojas et al., (2024) shows a collection of papers investigating different combinations of parameters, leading the authors to observe a lack of standardized methods to assess IAQ, along with a lack of long-term experimental studies. However, a study aimed to identify guidelines for specific VOCs rather than TVOC in the UK (Shrubsole et al. 2019) reported generally low concentrations of VOCs with health impact after few months after construction, possibly indicating that TVOC monitoring is adequate in residential settings. It also indicates that source control is the preferential strategy to mitigate their concentration and reduce exposure of the indoor population, along with good ventilation practices to dissipate residues. Considering current sensing technology, many studies involving pollutants and VOCs use passive samplers (Derbez et al. 2014a; Derbez et al. 2014b; Langer et al. 2015; Kauneliene et al. 2016; Piasecki 2019), which are adequate for experimental study but are not of practical use in a smart sensor network since they need to be periodically collected and analysed. In Calvo et al. (2022), TVOCs are measured with the previously mentioned CCS811 (Datasheet, CSS811 2020), which is based on a solid-state metal oxide semiconducting (MOX) gas-sensing element (Zampolli et al. 2005). It is very compact and low-power, has a range of 0-1187 ppb, is calibrated to a typical compound mixture for an indoor environment and features an automatic baseline correction with a 24-h period. Another gas sensor based on MOX technology is TGS2602 (Datasheet, TGS2602 2018), with a detection range of 130 ppm, as mentioned previously (Xiang et al. 2013). Furthermore, metal oxide semiconductors are an inexpensive option to monitor CO, O 3 and INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 19 NO x (Snyder et al. 2013). Finally, PM monitoring is most commonly performed through the light scattering technique, which is inexpensive and compact but does not provide direct mass measurements and can have issues with ultra-fine particles (Snyder et al. 2013). For example, the GP2Y1010AU0F sensor used in the work by Qabbal, Younsi, and Naji (2022) to monitor PM 2.5 is a low-end basic model produced by SHARP/Socle Technology (Datasheet GP2Y1010AU0F, 2018), with a declared measurement accuracy of 100 g m –3 ± 30%. However, despite the information reported in (Qabbal, Younsi, and Naji 2022), this sensor is unable to discern between PM 2.5 and PM 10 according to the manufacturer. Nevertheless, there are other components that have this ability (e.g. GP2Y1030/31AU0F detects PM 2.5 and PM 10 separately, and DN7C3CA007/DN7C3D015 is solely dedicated to PM 2.5 ).ata supporting the findings of this study are avai 5.2.2. Temperature and humidity One of the most relevant components of IEQ is TC. Although it is affected by several parameters, according to the well-known Fanger model (Fanger 1967), the most commonly monitored parameters are temperature and relative humidity. More precisely, this model is based on measurements of both air and mean radiant temperatures since it needs to evaluate simultaneously the convective and radiative heat exchange between the human body and the surrounding environment. In practice, the indoor TC is empirically assessed by measuring the air and operating temperatures. However, there are IEQ protocols, such as the recently published TAIL (Wargocki et al. 2021), that evaluate the thermal environment through the air temperature only. This simplification is based on the assumption that, in low-energy buildings, the difference between these two quantities is negligible; for instance, a study by Dawe et al. (2020) investigated the difference between air temperature and mean radiant temperature over 200k pairs of measurements from field and laboratory analyses and shows that the median absolute difference is 0.4 °C, regardless of the building type or the system type and operation, which is well below the temporal or spatial fluctuation. This result is also supported by the work of Kontes et al. (2017), which obtains a similar outcome for buildings with high thermal mass when solar and internal gains are properly taken into account, while more caution is required with buildings with lower thermal mass. As far as sensors are concerned, there are several options available on the market (Dong et al. 2019), with different levels of precision and accuracy, ranging from less than ± 0.5 C to around ± 0.1 C. To ensure stability and address any drift, they require regular calibration. A low-cost solution to measure both air temperature and relative humidity at the same time is the capacitive-type sensor DHT22 (Datasheet, DHT22 2012), which is used as part of a smart sensor to assess TC in a retrofitted university building in Lille (France) (Qabbal, Younsi, and Naji 2022). In this paper, it is declared to measure relative humidity and air temperature in the range of 0%…100% and −40…80 °C, respectively, with ± 5% and ± 2 °C accuracy. This same sensor is also used in few other works from the literature: first, a study by Mui et al. (2016), aimed at developing a predictive IEQ calculator, which is based on few sensors available on the market; second, an occupancy detection algorithm based on the sampling of several environmental variables and involving statistical learning models (Candanedo and Feldheim 2016); and, finally, a low-cost do-it-yourself device to monitor environmental parameters and control heating, cooling, ventilation and lighting presented in (Salamone et al. 2017). Even though it is nominally the same device, its specifications seem different in some aspects (i.e. relative humidity range and accuracy of 0…100% and ± 2% or ± 3%, respectively, and temperature accuracy of ± 0.5 °C). Importantly, the temperature accuracy reported in all the studies seems to be in disagreement with the value of ± 0.2 °C declared by the manufacturer in the data sheet (after factory calibration in a dedicated thermostatic chamber), and no explanation for this issue is provided in any case. For this reason, the data included in Table 4 refer to the data sheet provided by the manufacturer. Another commercially available device is the Telaire T9602 (Datasheet, T9602 2018), which includes a capacitive polymer relative humidity sensor ( ± 2% accuracy in the 2080% range) and a proportional to absolute temperature sensor ( ± 0.5 °C accuracy in the 2040 °C range). An example of its application can be found in (Weyers et al. 2017), where it is used to develop an IEQ platform dedicated to classrooms and based on low-cost components. According to the authors, this sensor has been selected for its reliability, and its connection to the main board of the Smart Sensor platform has been defined keeping in consideration its sensitivity to heat sources, since the heat radiated by the microcontroller or other sensors could interfere with the accuracy of data readings. A third option mentioned in the literature is the 20 A. ALONGI ET AL. BME680 sensor (BME 2024): it is used to monitor temperature and relative humidity to provide a realtime evaluation of the spatial and temporal distributions of PMV in a study by Mohammadi et al. (2024). In this work, the measurement range and accuracy declared are −4085 °C and ± 1 °C for temperature and 0…100% and ± 3% for relative humidity. Importantly, this is actually a very compact and low-power 4-in-1 sensor capable of monitoring gas concentration and pressure and, therefore, a potentially convenient solution for IoT devices. Among the all-in-one smart sensors (see § Section 5.2.6 for a detailed discussion), a noteworthy solution in terms of temperature and relative humidity measurement is the one described in the paper by Parkinson et al. (2019), called SAMBA. In detail, the housing of this device includes two main components: a main unit and a satellite element, which includes all sensors involved in local TC assessment: air and operative temperature (NTC thermistor, range 050 °C, resolution 0.1 °C) relative humidity (capacitive, range 595%, resolution 0.1%) and air velocity (bi-directional thermal anemometer, range 01 m/s, resolution 0.01 m/s). This solution is adopted after testing, instead of a single-housing unit, to address potential measurement biases due to waste heat from other components (e.g. power-conditioning circuits and other sensors). Finally, many studies related to IEQ in general and indoor TC specifically involve different kinds of monitoring station (Lan et al. 2011; Painter et al. 2012; Ekwevugbe et al. 2017; Devitofrancesco et al. 2019; Piasecki 2019). However, even though these clusters of sensors equipped with a dedicated data-logger are able to provide a complete environmental monitoring solution, they are intended for research and experimental use and do not present any interest in everyday domestic applications in smart sensor networks. 5.2.3. Light As stated in Section 3.2, VC is an important component of IEQ that directly affects occupant productivity, especially in office buildings. When integrated in an IoT network, light sensors (i.e. photometric sensors) can be used to control the activation and intensity of luminaires according to daylight availability to reduce electricity consumption while ensuring human comfort (Yan et al. 2017; Navada et al. 2013; Heidary et al. 2023). However, the effectiveness of this approach largely depends on the placement of these sensors in the built environment in relation to the locations typically occupied in indoor spaces (Heidary et al. 2023; Dong et al. 2019). Moreover, even though VC can be based only on illuminance and daylight factors (Wargocki et al. 2021), other aspects can be taken into consideration in the control strategy of the lighting system, such as the colour temperature ratio, glare, light spectrum, etc. (Dong et al. 2019). An example of sensor-based lighting control is presented in (Pandharipande and Caicedo 2015), where photometric sensors are used in combination with occupancy sensors in a distributed architecture to detect human presence and control IoT-ready luminaires to guarantee 500 lux on used working stations and 300 lux everywhere else. The system, which requires calibration against the dimming levels of the luminaires, is controlled through an optimised algorithm, which shows promising results when compared to a simple PID control but is largely affected by the unknown effects of daylight on the sensor planes (i.e. available daylight on the working plane might differ from the one perceived by the sensor on the ceiling due to unmapped reflections). This work also highlights several technical challenges: first, due to the importance of daylight, the control algorithm needs to be able to account for blinds and movable shading devices; second, the control system must consider the effects of daylight penetration in the indoor space and the lighting design (i.e. luminaire location, spacing, etc.); finally, user satisfaction needs to be monitored and considered in the control strategy. Light sensors are also used in some cases to provide indications of occupancy in rooms (Candanedo and Feldheim 2016). A low-cost photometric sensor available on the market is the TSL2561 (Datasheet, TSL2561 2025), which is based on a photodiode and is mentioned in some works in the literature (Candanedo and Feldheim 2016; Mui et al. 2016): it has a range of 0.140000 lux and is able to detect infrared and fullspectrum light separately. Although it is a do-it-yourself option, it performs comparably to laboratorygrade instruments when involved in VC assessment (Mui et al. 2016). INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 21 5.2.4. Acoustics The sound pressure level is generally used in rating schemes to address indoor AC in buildings (Devitofrancesco et al. 2019; Wargocki et al. 2021; Rocca et al. 2022), since it has been correlated with subjectively perceived noise discomfort (Huang and Griffin 2012) and subsequent performance in a work setting due to its impact on speech intelligibility (Hongisto 2005) (which is evaluated through the dedicated STI (Devitofrancesco et al. 2019). Moreover, as mentioned in § Section 3.2, it has been demonstrated that ambient noise in a residential context can disrupt sleep quality, inducing tiredness and reducing daytime performance, with a detrimental effect on overall health (Muzet 2007). However, despite its significance, not many works in the literature involve indoor AC measurement, in comparison to the other components of IEQ: in some cases, the sampling campaign is based on laboratorygrade equipment (Devitofrancesco et al. 2019; Lan et al. 2011; Huang and Griffin 2012; Piasecki 2019), which would not be useful in an IoT network inside a residential Smart Building, while in other manuscripts, the sound pressure levels are used only as a mean to estimate occupancy levels (Ekwevugbe et al. 2017; Zikos et al. 2016; Dong and Andrews 2009), rather than to evaluate AC itself. On the other hand, some studies involve compact and commercially available low-cost sensors that can be used in an IEQ-oriented smart sensor: one example is the Grove Sound Sensor (Datasheet, Grove 2015), which includes an electret microphone and an LM358 amplifier, and generates an analog output. This device is used in studies that adopt a do-it-yourself approach: For instance, Mui et al. (2016) described a smart sensor (IEQ calculator in the paper) based on compact low-cost sensors and compared its accuracy against laboratory-grade equipment by correlating the two IEQ calculations in an office environment. However, among all the measurements, which generally show high accuracy and sensitivity (i.e. R 2 > 0.95, p > 0.7), the sound pressure level is characterized by a worse performance (i.e. R 2 > 0.65, p > 0.45), possibly due to background noise fluctuations. Nevertheless, the authors still consider it acceptable, due to the small contribution of AC to the overall IEQ in their model. An electret microphone is also adopted in other cases: first, in the bespoke PCB described in (Parkinson et al. 2019), to measure the sound pressure level (range 4090 dBA, resolution 0.1 dBA). Second, in the work of Zikos et al. (2016), it is used to register acoustic events that are tied to human presence and occupancy level, with a high degree of efficiency in terms of both accuracy and protection of privacy. Finally, this type of device is used in the Smart Sensor described in (Qabbal, Younsi, and Naji 2022), but results about AC are not presented in the manuscript. 5.2.5. Occupancy Although occupancy monitoring is not strictly related to IEQ assessment, significant research effort has been dedicated to this topic in recent years. Indeed, information about occupancy (i.e. human presence, number of occupants, type of activity, location in space, identification) can be integrated into the BMS and facilitate improvements in energy efficiency and personalised environmental comfort (Jung and Jazizadeh 2019; Wang et al. 2005; Lam et al. 2014; Pandharipande and Caicedo 2015; Zikos et al. 2016; Chen et al. 2016; Candanedo and Feldheim 2016; Shrubsole et al. 2019; Kim and Srebric 2017; Sun et al. 2020): as an example, a control strategy for HVAC systems based on room occupancy prediction is presented in (Erickson and Cerpa 2010), which leads to 20% potential energy savings. A similar approach is the subject of (von Bomhard et al. 2016), where measurement-based occupancy prediction is used to control the activation of the heating system in each room separately. The work by Capozzoli et al. (2017) involves the definition of an occupancy-based energy saving strategy for the HVAC system, which leads to 14% average energy savings during the investigated period. In (Jia et al. 2017), the VAV boxes are controlled according to an occupancy-location model, with a specific focus on privacy protection. In fact, a significant portion of the research effort in the last year has been dedicated to the definition of reliable predictive algorithms for occupancy, which are based on direct or indirect measurements (Wang et al. 2018; Jia et al. 2017; Kleiminger et al. 2014; Zuraimi et al. 2017), which can also be used in building numerical simulation and the definition of digital twins (Capozzoli et al. 2017; Page et al. 2008; Duarte et al. 2013; D’Oca and Hong 2015; Mahdavi and Tahmasebi 2015; Davis and Nutter 2010; Chen et al. 2015; Yan et al. 2017). The list of all sensors and techniques mentioned in this section is summarized in Table 5. 22 A. ALONGI ET AL. In terms of direct measurement, the review by Dong et al. (2019) indicates the following categories of sensors: image-based sensors, motion sensors, radio-based sensors and threshold and mechanical sensors. The first group, which includes infrared, visible light and luminance cameras, is generally adopted in office buildings for security purposes (Benezeth et al. 2011; Shih 2014; Labeodan et al. 2015; Zou et al. 2017) and can be used to track occupants or identify their location, number, activity and identity. The main limitations of this approach, as confirmed in (Sun et al. 2020), are due to its cost, its limited coverage caused by occlusions, the complexity of the algorithms required and the concerns due to privacy infringement (Jiang et al. 2016). This last issue can be addressed by using depth sensors instead of cameras, since it has been demonstrated that depth data can count and locate occupants, even in crowded environments, but are not suitable for personal identification, hence preserving privacy (Galcık and Gargalik 2013; Seer et al. 2014; Diraco et al. 2015) when no RGB camera is involved. Motion sensors (i.e. PIR sensors, ultrasonic Doppler, mocrowave Doppler, photosensors, etc.) are generally adopted to control artificial lighting and HVAC for energy saving purposes. These devices are effective in detecting occupancy presence, but are unable to count subjects or detect small movements, thus providing a false control signal when occupants are not moving for a long time (Heidary et al. 2023; Akbar et al. 2015; Guo et al. 2010; Song et al. 2014), leading to potential discomfort or increased energy consumption. PIR sensors, which measure changes in infrared light emitted by objects in their field of view, are cost-effective options that are frequently used in non-residential settings to detect occupant presence (Ekwevugbe et al. 2017; Weyers et al. 2017). More precisely, since these sensors detect movements, which are discrete events, a time delay value (e.g. 15–60 min) is defined to avoid misreading vacancies due to immobility, and the space is considered to be occupied during this period. The appropriate setting of this parameter is necessary to avoid false absence detection (i.e. underestimation of the delay) or a reduction in potential energy savings (i.e. overestimation of the delay) (Shen et al. 2017). Table 5. List of sensors and techniques to detect occupancy-related data in buildings. The works from literature where they are discussed are reported in column Refs. Method Technology References Notes Direct Image-based sensor Infrared, visible light and luminance cameras (Zou et al. 2017) (Shih 2014) (Sun et al. 2020) (Benezeth et al. 2011) (Labeodan et al. 2015) (Jiang et al. 2016) It can track occupants location, number, activity and identity it is costly, limited by occlusions, requires storage and computing power, presents privacy issues Direct Image-based sensor Depth sensor (Galcık and Gargalik 2013) (Seer et al. 2014) (Diraco et al. 2015) It can count and locate occupant without identification (privacy preservation) Direct Motion sensor PIR (Ekwevugbe et al. 2017) (Weyers et al. 2017) (Labeodan et al. 2015) (Shen et al. 2017) (Wahl et al. 2012) (Raykov et al. 2016) Assesses occupancy by detecting movements and requires a time delay to avoid misreading vacancy, people counting can be performed with appropriate configurations Direct Radio-based sensor RFID (Raykov et al. 2016) (Li et al. 2012) (Ranjan et al. 2013) Identifies occupants through radio tags with a univocal identification code (terminal based) Direct Threshold/ mechanical sensor Reed switches, door badges, piezoelectric mats or infrared beams (Agarwal et al. 2010) Interaction with openings is required for the system to function Indirect Environmental variable CO 2 concentration (Qabbal et al. 2022) (Calì et al. 2015) (Zuraimi et al. 2017) (Jiang et al. 2016) Requires machine learning or predictive algorithm, observed accuracy above 80%, it can have issues since CO 2 is a transient phenomenon Direct Combined measurements CO 2 concentration and PIR (Akkaya et al. 2015) (Gruber et al. 2014) Improved accuracy in estimating occupancy level Indirect Combined measurements CO 2 concentration and other IEQ variables (Dong and Andrews 2009) (Zikos et al. 2016) (Dong et al. 2010) (Chen et al. 2017) Improved accuracy in estimating occupancy level Indirect Other techniques Webcams, mouse and keyboard activity, chair sensors, Wi-Fi connection, electricity consumption (Newsham et al. 2017) (Zhao et al. 2015) (Zou et al. 2017) (Labeodan et al. 2015) (Zou et al. 2018; Becker and Kleiminger 2018) Potential privacy issues Direct Motion sensor mmWave (Gu et al. 2019) (Pegoraro et al. 2022) (Liu et al. 2024) (Zhang et al. 2025) (Haipeng et al. 2021) (Zhang et al. 2023) Effective in localizing multiple people simultaneously, sensitive to small movements, privacy-preserving INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 23 However, it has been shown that a false ON state can be triggered by warm currents from HVAC systems (Labeodan et al. 2015). Finally, the work of Wahl et al. (Wahl et al. 2012) demonstrated that a PIR sensing network can also be used to estimate people count, by installing sensors in pair and supporting them with one of two distributed algorithms proposed in the paper, which are defined to perform the count estimation using the directional information from the sensor pairs, possibly including the sensor masking time. In (Raykov et al. 2016), a single PIR sensor combined with an ML model was shown to be capable of performing this same task, even if at a lower accuracy than more state-of-the-art techniques. Radio-based sensors can detect occupant presence using radio signals. For instance, RFID can sense people using tags with a univocal identification code and therefore can be considered a terminal-based detection system (Wu and Wang 2019). In (Li et al. 2012), where this technology is used to develop an occupancy detection system that can be used to run demand-driven HVAC operations, RFID is considered a valuable alternative to image-based and motion sensors since it is cost effective, adequately accurate, and able to detect multiple stationary subjects while not requiring a line of sight and on-board storage. In their work, Ranjan et al. (Ranjan et al. 2013) introduced the so-called RF Doormat system, which includes two separate RF sensing zones (one on each side of a door), RFID ankle bracelets with passive tags (i.e. no battery requirement) and a door-crossing detection algorithm. This system is tested in a residential setting as a way to infer people's room location and has an accuracy of 98%. A similar approach is presented in the work of Hnat et al. (Hnat et al. 2012), where the authors use ultrasonic range-finding sensors as primary devices at the doorway. As far as threshold and mechanical sensors are concerned, they are able to record occupants' presence in an enclosed environment through their interactions with either windows or doors, and they can be reed switches, door badges, piezoelectric mats or infrared beams. These systems can either require direct interaction with the occupants (i.e. door badges) or operate autonomously, but in all cases, they are prone to errors when counting occupants (e.g. multiple people crossing an infrared beam at the same time). A reed switch is used in the work of Agarwal et al. (Agarwal et al. 2010) in combination with a PIR motion detection device to detect occupancy in a single room office. This synergy presence node design uses the signal from both sensors in a dedicated algorithm that allows for a high degree of accuracy and, according to the authors, could lead to 10%–15% potential energy savings with the introduction of an occupancydriven HVAC control strategy. Along with direct occupancy measurements, which present several issues (invasiveness, cost and complexity of the hardware, installation requirements, privacy intrusion) despite good accuracy, significant research effort is dedicated to indirect measurement techniques that infer the occupancy level by analysing its effect on environmental variables that are already monitored for IEQ (Viani et al. 2014). For instance, in (Qabbal et al. 2022), CO 2 is considered a proxy for occupant density in indoor environments: its concentration in occupied classrooms increases significantly, with a strong correlation with the number of people. Jiang et al. (Jiang et al. 2016) described an ML algorithm that processes smoothed CO 2 measurements and counts the occupants, showing accuracies between 89% and 94% when tested in a large office with up to 35 occupants (assuming a tolerance of 4 people). A comparable outcome is shown in (Zuraimi et al. 2017), where CO 2 readings combined with dedicated prediction models (physical or statistical) are used to estimate the number of occupants in a lecture theatre. However, according to the review by Shen et al. (Shen et al. 2017), zone-level occupancy estimation based on CO 2 readings can be challenging since CO 2 transport is an intrinsically transient phenomenon and can be influenced by the furniture layout and the distance between the sensor and the occupants, as demonstrated in (Calì et al. 2015). For this reason, several works suggest to combine CO 2 readings with other sensing techniques, such as PIR (Akkaya et al. 2015; Gruber et al. 2014) or other environmental variables (Dong and Andrews 2009; Zikos et al. 2016; Dong et al. 2010; Chen et al. 2017), to increase accuracy. Finally, other manuscripts, which are more focused on office buildings, suggest other means to collect occupancy information, such as webcams, mouse and keyboard activity, chair sensors, Wi-Fi connections, electricity consumption (Newsham et al. 2017; Zhao et al. 2015; Zou et al. 2017; Zou et al. 2018; Labeodan et al. 2015; Becker and Kleiminger 2018). Finally, in recent years, much interest has been gathered by the so-called mmWave, which can be used not only for effective wireless communication but also for privacy-preserving sensing technology to detect human presence in smart buildings, since it is effective in localising multiple people simultaneously 24 A. ALONGI ET AL. occupant spaces where individual preferences vary. Indeed, studies show that while AI-driven personal comfort models (Kim et al. 2018; Li et al. 2017) can adapt to individual needs, they struggle to reconcile conflicting preferences in shared environments (Ghahramani et al. 2014). New solutions like the smart token-based scheduling algorithm (Png et al. 2019) use decentralised designs to speed up systems, and augmented reality interfaces (Mohammadi et al. 2024) provide a new way to engage with them. However, a comparative study (Murakami et al. 2007) shows that users prefer passive systems since they are less likely to cause problems. Indeed, one of the biggest problems with human-inthe-loop systems is finding a balance between privacy (keeping data collection from being intrusive) and usability (making it easy for users to use). Studies offer a number of alternatives, such as federated learning, where decentralised AI models digest data on user devices and only send back anonymised information for HVAC optimisation (Li et al. 2017). Another way to understand how users behave over time is to use automated occupant profiling with ML algorithms, reducing the need for frequent manual changes (Behzadi et al. 2023). In another study, Peng et al. (Peng et al. 2017) used binary presence detection to change the temperature set-points, depending on an estimate of how many people will be present at that moment. In addition, the use of CO 2 and motion sensors that do not require any physical interaction provides occupancy data without identification, hence preserving privacy, but might not be detailed enough for personalised comfort (Kim and Srebric 2017). Despite these advances, privacy concerns persist, particularly in workplaces and shared residential spaces. The collective decision algorithm (Li et al. 2017) proposes to solve this problem by combining anonymous comments from many people to make the group more comfortable. However, it still requires an opt-in agreement. Another study (Zhao et al. 2015) used anonymised data, which improved the detection accuracy from 62% to 95%. 6.4. Case studies The case studies revealed that the technological landscape of the IEQ-oriented BMS is characterised by advanced sensor networks, predictive modelling and adaptive control systems (Cascone et al. 2017 ; Erickson and Cerpa 2012; Dong et al. 2010). Real-time monitoring is also well spread, from low-cost sensor networks (Weyers et al. 2017) (e.g. SKOMOBO for monitoring IAQ in classrooms) to large and dense sensor architectures (Malkawi et al. 2023) (e.g. HouseZero's temperature, humidity, and CO 2 sensors), with the latter integrating more than 300 sensors for the control of both HVAC and lighting across hybrid control layers. Occupancy detection is also a significant driver, using either direct measurements or indirect CO 2 -based algorithms (Calì et al. 2015,Jiang et al. 2016,Lu et al. 2011), 3D depth sensors (Galcık and Gargalik 2013; Seer et al. 2014; Diraco et al. 2015), and occupancy patterns deduced from WiFi (Zou et al. 2017; Zou et al. 2018). Predictive HVAC control is a key feature in the smart building prototype, including hybrid CNN-RNN models for the prediction of energy needs (Sharma et al. 2024) and model predictive control for the optimisation of TC (Bengea et al. 2015; Goyal et al. 2015). When it is incorporated within BMS for energy management, as in federal Canadian buildings, it can achieve energy savings of up to 15% by means of anomaly detection (Shen et al. 2017). A study on comfort modeling investigated techniques ranging from PMV TC (Anselmi and Moriyama 2017) to participatory frameworks, such as Thermovote, which defines HVAC set points through occupant voting (Salamone et al. 2018). However, there are discrepancies in sensor accuracy (e.g. CO 2 sensors' limitation in regional applications (Shen et al. 2017; Calvo et al. 2022) and prevalence (e.g. RFID-based system infeasibility for large environments (Li et al. 2012; Behzadi et al. 2023; Kofler et al. 2012) and privacy (Wang et al. 2020; Kofler et al. 2012). Energy efficiency and comfort usually compete when IEQ factors are integrated within BMS logic. For example, DCV paradigms are energy saving, but they can underperform in steady-state conditions, potentially leading to pollutant accumulation (Lu et al. 2011;O’Neill et al. 2020), and personalised comfort systems (e.g. smartphone-controlled HVAC (Li et al. 2017; Ortiz et al. 2020)) are difficult to operate in multi-occupant spaces. Adaptive granularity is also found in lighting systems, where luminare-based sensors can save 30.8% (Feyzi and Mojallali 2024) but suffer computational burdens in centralised designs (Pandharipande and Caicedo 2015). The dependence on real-world data, as explained in the Danish residential heat meter case study (Sharma et al. 2024), increases validity but reveals discrepancies in INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 31 applicability across different climates. Studies such as (Li et al. 2019; Shen et al. 2017; Zhao et al. 2015; Kofler et al. 2012) attempt to balance energy efficiency and occupant comfort by performing dense sensor deployment to record parameters such as CO -3 , temperature and occupancy. For example, at Hotel ICON (Hong Kong Li et al. 2019), a three-step BMS cycle (monitoring, diagnosis, intervention) is tested to strike for performance-based optimization of the operation of the fan coil unit, according to the comfort needs of the occupants and the situational operating patterns. Other methods use motion sensors (Peng et al. 2017), dense sensor networks (Wang et al. 2020), CO 2 data (Jiang et al. 2016), human feedback (Erickson and Cerpa 2012), and hybrid deep learning techniques (Sharma et al. 2024) to adapt HVAC and lighting dynamically. However, the manner in which IEQ parameters are integrated into the BMS logic is highly variable. Some case studies prioritize TC through PMV indices (Anselmi and Moriyama 2017), IAQ via the CO 2 concentration (Choi and Yeom 2017) or visual comfort via adaptive lighting (Pandharipande and Caicedo 2015). This variability demonstrates that there is no one-size-fits-all approach to the integration of IEQ, and the outcomes are largely determined by context, including the type of building and the pattern of use. Sensor accuracy can sometimes present issues: even though the performance of low-cost IAQ sensors can be adequate when compared to recommended readings and be sufficient to assess the overall condition of the indoor climate while providing great scalability (Weyers et al. 2017), CO 2 -based occupancy predictors used to control ventilation can perform poorly in the presence of high airflow rates (Gruber et al. 2014), especially when the sensor is located inside the room and not in the exhaust air duct (higher measurement fluctuations lead to inconsistent behaviour). Ethical issues, such as privacy risks of IoT occupancy tracking (Cascone et al. 2017), are often not adequately examined, and limited research has been proposed to mitigate them (e.g. lightweight cryptography). Occupancy prediction models (e.g. KNN algorithms (Peng et al. 2017)) fail on unusual patterns, while TC models such as PMV often do not agree adequately with subjective feedback, especially in residential environments (Becker and Paciuk 2009). Scalability represents another challenge (Kofler et al. 2012): on the one hand, centralised systems (e.g. Hotel ICON's triadic BMS Li et al. 2019) require large capital, on the other hand, decentralised solutions (e.g. do-it-yourself IoT solutions Salamone et al. 2017) suffer from interoperability issues. Results discrepancies add additional complexity to the subject. For example, the measured occupancybased optimal (MOBO) controllers obtained 40% energy savings in HVAC simulations but only 30% savings in field empirical tests because the logic of the AHU was not ideal (Goyal et al. 2015). In addition to performance indicators, discrepancies also occur between official environmental quality certification metrics and user satisfaction. As discussed in § 3.3, many LEED-certified buildings do not achieve high occupant satisfaction despite accomplishing high IEQ scores (Asmar, Chokor, and Srour 2014; Altomonte et al. 2019). This ignores cultural and contextual factors: wrist skin temperature models (Dong and Andrews 2009; Sim et al. 2016) may not generalise to all demographics, and daytime harvesting strategies are highly dependent on the geographical orientation of users (Navada, Adiga, and Kini 2013). These incongruities highlight the importance of adaptive and context-sensitive BMS in mediating occupant preferences while respecting technical feasibility and energy efficiency. To overcome these limitations, future applications should focus on adaptive learning and occupantcentred reconfigurability. Participatory systems (e.g. Thermovote (Erickson and Cerpa 2012) and OPTC frameworks (Lam et al. 2014)) show the possibilities of human-in-the-loop control, but such systems clearly need yet another coherent feedback mechanism to not overwhelm the user. Hybrid methods, such as federated learning for personalised comfort models (Ghahramani et al. 2015), may alleviate the issue of data privacy and improve accuracy. Scalability can be provided by modular BMS architectures, such as the KNX-BACnet with hybrid system (Malkawi et al. 2023), assuming that interoperability standards (e.g. in IoT protocols (Calvo et al. 2022)) are defined and consistently implemented. 7. Challenges and future developments The analysis of the literature demonstrates that the assessment and management of IEQ in smart buildings is a multidisciplinary endeavour that must be tackled at different levels and from multiple perspectives, presenting challenges that need to be addressed in future research developments. 32 A. ALONGI ET AL. First, standards and schemes addressing IEQ in buildings require further investigation, due to their relevance for both the design phase and the management phase throughout a building life cycle, especially in residential settings. On the one hand, there is currently no consensus on the weighting of each IEQ component in the overall performance, since it depends on several factors, ultimately suggesting that a single index may not always be required and each component could be considered independently, as in the TAIL protocol. Moreover, additional effort is needed to investigate further the combined impact of different IEQ factors and their effects on occupant health and well-being. This approach provides a robust set of tools to support technicians during the design phase. On the other hand, it has been observed that careful design does not always align with occupants’ perceptions of IEQ during the use of the building, especially when statistically-based schemes are adopted for small groups, such as in residential settings. Therefore, it is essential to recognise the needs of individual occupants and further develop personalised comfort models that can be deployed during the use phase of a building to assess perceived comfort levels. This would also enable the consideration of several parameters currently neglected by standards, such as different regional, cultural, climatic and building characteristics, and the definition of specific scoring systems, while maintaining a common base framework. Second, a crucial factor to achieve high levels of perceived IEQ is the adoption of a well-designed network of smart sensors, which also needs a high degree of flexibility and scalability to accommodate changes in requirements throughout the lifetime of a building. One of the most critical challenges is the effect of the spatial distribution of environmental variables on the accuracy of the monitoring system. This issue should be considered when designing the sensor network in a smart building and can be more prevalent in residential contexts where the disposition of the furniture, and therefore that of the occupants, can be less predictable than in office environments. Further research on this subject could result in guidelines to support this design process. Considering the technology of Smart Sensors themselves, future efforts should focus on further improving their accuracy, since it significantly affects the quality and reliability of the collected data and therefore the efficacy of the BMS. Furthermore, the selection of sensing elements should be limited to the most relevant environmental variables for the specific building category to reduce the energy consumption of the network and the general installation cost, thus facilitating their diffusion in the market. If we focus on the main IEQ components, the following considerations arise: a. There is a lack of consistency in the monitoring strategies for IAQ. Past works suggest that in residential buildings VOCs typically fall below potentially hazardous concentrations only a few months after construction and that simple source control can be an effective strategy, along with direct measurement of CO 2 concentration and TVOC sampling. In this context, an adequate air change rate and effective PMs filtration can maintain high levels of IAQ. b. To evaluate TC, the air temperature is generally accepted as a substitute for the operating temperature, especially when highly insulated buildings are considered. This approach can therefore simplify Smart Sensor design but may require adjustments in cases of significant radiative asymmetry (e.g. next to windows). c. When assessing VC, studies have shown that the efficacy of light sensors is greatly affected by internal reflections, which should be mapped to allow the system to perform correctly. Moreover, the BMS needs to be able to manage both artificial and natural light to guarantee a comfortable environment from the visual standpoint, which requires integration into the IoT network of actuators to control shading devices. d. Experimental studies involving AC evaluation are less prevalent in the literature than those dedicated to other IEQ components, showing the need for further investigations, especially in residential settings. The sound pressure level is generally sufficient to characterise AC, and low-cost sensors are available on the market. However, their accuracy seems to be significantly worse than that of lab-grade instruments, especially in the presence of background noises, and should be a focus of future research. Even though not required by any IEQ model, occupancy monitoring is a useful tool to support the BMS in a smart building in improving energy performance and guaranteeing adequate indoor conditions. However, it constitutes an intrusion of personal privacy, especially in residential applications. On this topic, mmWave technology seems to be a promising solution, both in terms of performance and privacy, and in the future, it could also provide a way for occupants to interact with the IoT through gestures. Alternatively, environmental variable monitoring (e.g. CO2 concentration and acoustic events) can serve as a meaningful proxy for occupancy evaluation, balancing system effectiveness and privacy for users. INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 33 However, privacy concerns extend to all data collected by the IoT network, especially when they are sent off-site to a cloud computing layer. This creates a significant vulnerability that requires further development. Finally, the effective operation of a smart building is largely based on its BMS, and recent research trends include the adoption of AI and ML techniques. The accuracy of these approaches is tied to the complexity of the models, along with the reliability of the smart sensor network. However, further developments should strive to find a trade-off between model accuracy and practicality, since complex models can be too computationally intensive for real-time implementation, despite their accuracy, while simpler ones can be easier to implement but might compromise user comfort due to lower precision. Moreover, another limitation of AI and ML algorithms is that they are trained on limited or contextspecific datasets, reducing their ability to generalise across different building types and, therefore, limiting their potential diffusion in the market. Finally, the efficacy of the BMS in maintaining adequate IEQ and satisfying occupant requirements also relies on the ability to collect their feedback. However, it can be challenging to engage user to actively provide feedback, potentially resulting in low adoption rates due to fatigue. A potential solution to address this issue could rely on the adoption of wearable sensors (e.g. smart watches, fitness bands or rings) to monitor physiological responses and assess comfort levels without requiring active user participation. This could also enable the ML algorithms to effectively predict occupant behaviour and their interactions with the surrounding environment, leading to a clear transition from reactive to proactive and context-aware building management strategies. On this note, occupants may also need the support of technicians in the training process of the autonomic BMS to make it more effective. 8. Conclusions This review investigates the literature dedicated to IEQ assessment and management in smart buildings through IoT networks and smart sensors, with a special focus on residential buildings. The list of references includes the European and US technical standards currently in force, along with the research papers published since 2010 (with only a few exceptions). Due to the broad scope of this work, it has been possible to observe several trends that have developed in the research landscape in recent years. Significant effort has been dedicated to the definition of IEQ models, for instance, with the introduction of personalised comfort models. Studies on smart sensors are focused more on environmental measurement issues (i.e. the effects of spatial and temporal fluctuations) and their integration with BMS, rather than on the sensing elements themselves. The only exception is occupancy sensors, which have been the subject of several manuscripts investigating various approaches. Finally, another fast-developing field of research is related to BMS, and the implementation of ML techniques and AI. Although BMS has been demonstrated to be an integral part of the functionality of smart buildings, challenges remain to be overcome, such as measurement accuracy due to sensor placement, privacy and system integration. Whether it is AI or IoT, the fast-paced development in technology provides a silver lining to these issues, further indicating that computerization in BMS, for a future that is both effective and occupant-centric, is the correct path. This review of the literature has shown that the IEQ in smart buildings is a multidisciplinary topic that can be addressed from both the engineering side and the occupant perspective. This means that future developments will require a multidisciplinary approach involving predictive models that can consider occupant preferences; minimal smart sensor and IoT networks that optimise the balance between affordability, efficacy and usability; a greater focus on data safety and privacy protection; and, finally, improved and frictionless interactions between autonomic BMS and users, potentially through the combined use of AI and unobtrusive wearable sensors, which would use occupant feedback to enable smart buildings to behave proactively in the pursuit of satisfactory IEQ and energy efficiency. It is evident that sensorand IoT-based IEQ management systems are highly dependent on an uninterrupted power supply and connectivity. With increasing climatic exacerbations and potential energy disruptions, future developments must therefore prioritise the resilience of smart sensor networks, ensuring adequate IEQ and occupant safety during temporary failures. Nevertheless, as extreme events become more frequent, 34 A. ALONGI ET AL. designers should still be aware of passive design strategies that can help to sustain acceptable IEQ conditions when active technological systems are compromised. Authors contributions CRediT: Andrea Alongi: Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing; Luca Pacileo: Investigation, Writing – original draft; Mustafa Muthanna Najm Shahrabani: Investigation, Writing – original draft; Paulius Spudys: Supervision, Writing – review & editing; Rossano Scoccia: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review & editing; Livio Livio Mazzarella: Conceptualization, Funding acquisition, Supervision. All authors revised it critically for intellectual content and the final approval of the version to be published. All authors agree to be accountable for all aspects of the work. Disclosure statement No potential conflict of interest was reported by the author(s). Funding This work was supported by the European Commission under the project ‘Boosting Research for a Smart and Carbon Neutral Built Environment with Digital Twins (SmartWins)’ (Grant ID Number 101078997), funded under the Horizon Europe call HORIZON-WIDERA-2021-ACCESS-03-01. Data availability statement The authors confirm that the data supporting the findings of this study are available within the article. Acronyms AC acoustic comfort AHU air handling unit AI artificial Intelligence BMI body Mass Index BMS building management system CNN building management system DCV demand controlled ventilation EPBD energy performance of building directive HVAC heating, ventilation and air conditioning IAP indoor air pollution IAQ indoor air qualityu IEQ indoor enviromental quality IoT internet of things LEED leadership in energy and environmental design ML machine learning mmWave millimiter-wave NDIR non-dispersive infrared PIR passive infra-red PM particulate matter PMV predicted mean vote POE post-occupancy evaluation POM passive observational method PPD predicted percentage of dissatisfied RFID radio frequency identification RL reinforced learning RNN recurrent neural network SBS Sick Building Syndrome SRI smart readiness indicator INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 35 STI speech transmission index TC thermal comfort TVOC total volatile organic compound VAV variable air volume VC visual comfort VOC volatile organic compound References Agarwal, Y., B. Balaji, R. Gupta, J. Lyles, M. Wei, and T. Weng. 2010. “Occupancy-Driven Energy Management for Smart Building Automation.” In Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building ACM, 1–6. doi: 10.1145/1878431.1878433. Akbar, A., M. Nati, F. Carrez, and K. Moessner. 2015. “Contextual Occupancy Detection for Smart Office by Pattern Recognition of Electricity Consumption Data.” In 2015 IEEE International Conference on Communications (ICC),561–566., London (United Kingdom): IEEE. doi: 10.1109/ICC.2015.7248381. Akkaya, K., I. Guvenc, R. Aygun, N. Pala, and A. Kadri. 2015. “IoT-Based Occupancy Monitoring Techniques for Energy-efficient Smart Buildings." In 1506 2015 IEEE Wireless Communications and Networking ConferenceWorkshops (WCNCW), 58–63. New Orleans (USA): IEEE. doi: 10.1109/WCNCW.2015.7122529. Altomonte, S., J. Allen, P.M. Bluyssen, G. Brager, L. Heschong, A. Loder, S. Schiavon, J. A. Veitch, L. Wang, and P. Wargocki. 2020. “Ten Questions Concerning Well-being in the Built Environment.” Building and Environment 180. doi: 10.1016/j.buildenv.2020.106949. Altomonte, S., S. Schiavon, M.G. Kent, and G. Brager. 2019. “Indoor Environmental Quality and Occupant Satisfaction in Green-certified Buildings.” Building Research Information 47(3): 255–274. doi: 10.1080/ 09613218.2018.1383715. Amasyali, K., and N.M. El-Gohary. 2016. “Energy-Related Values and Satisfaction Levels of Residential and Office Building Occupants.” Building and Environment 95: 251–263. doi: 10.1016/j.buildenv.2015.08.005. Amphenol Sensors – Telaire[online]. Accessed 08/04/2025. https://www.amphenol-sensors.com/en/telaire/co2. Anselmi, N., and T. Moriyama. 2017. “An Inversion Strategy for Energy Saving in Smart Building Through Wireless Monitoring.” Journal of Physics: Conference Series 904: 012003. doi: 10.1088/1742-6596/904/1/012003. Asadi, I., N. Mahyuddin, and P. Shafigh. 2017. “A Review on Indoor Environmental Quality (Ieq) And Energy Consumption in Building Based on Occupant Behavior.” Facilities 35: 684–695. doi: 10.1108/F-06-2016-0062. Asmar, M.E., A. Chokor, and I. Srour. 2014. “Occupant satisfaction with indoor environmental quality: A study of the LEED-certified buildings on the Arizona State University campus.” In ICSI 2014, American Society of Civil Engineers, 1063–1070. doi: 10.1061/9780784478745.100. Bayart, M. 2001. “LARII: Development Tool for Smart Sensors and Actuators.” In IFAC Proceedings, 34, 70–75. doi: 10.1016/S1474-6670(17)32795-7. Becker, R., and M. Paciuk. 2009. “Thermal Comfort in Residential Buildings–failure to Predict by Standard Model.” Building and Environment 44: 948–960. doi: 10.1016/j.buildenv.2008.06.011. Becker, V., and W. Kleiminger. 2018. “Exploring Zero-training Algorithms for Occupancy Detection Based on Smart Meter Measurements.” Computer Science – Research and Development 33: 25–36. doi: 10.1007/s00450-0170344-9. Behzadi, A., E. Thorin, C. Duwig, and S. Sadrizadeh. 2023. “Supply-Demand Side Management ofAsmar a Building Energy System Driven By Solar And Biomass in Stockholm: A Smart Integration With Minimal Cost And Emission.” Energy Conversion and Management 292: 117420. doi: 10.1016/j.enconman.2023.117420. Benezeth, Y., H. Laurent, B. Emile, and C. Rosenberger. 2011. “Towards a Sensor for Detecting Human Presence and Characterizing Activity.” Energy and Buildings 43: 305–314. doi: 10.1016/j.enbuild.2010.09.014. Bengea, S. C., P. Li, S. Sarkar, S. Vichik, V. Adetola, K. Kang, T. Lovett, F. Leonardi, and A. D. Kelman. 2015. “FaultTolerant Optimal Control of a Building HVAC system.” Science and Technology for the Built Environment 21(6): 734–751. doi: 10.1080/23744731.2015.1057085. Berger, C., A. Mahdavi, E. Azar, K. Bandurski, L. Bourikas, T. Harputlugil, R. Hellwig, R. Rupp, and M. Schweiker. 2022. “Reflections on the Evidentiary Basis of Indoor Air Quality Standards.” Energies 15: 7727. doi: 10.3390/ en15207727. von Bomhard, T., D. Wörner, and M. Röschlin. 2016. “Towards Smart Individual-room Heating for Residential Buildings.” Computer Science – Research and Development 31: 127–134. doi: 10.1007/s00450-014-0282-8. Bordass, W., A. Leaman, and J. Eley. 2006. A Guide to Feedback - A Guide to Feedback and Post-occupancy Evaluation. Tech. Rep. The Usable Building Trust. https://www.usablebuildings.co.uk/UsableBuildings/ Unprotected/AGuideToFeedbackAndPostOccupancyEvaluation.pdf. Bouktif, S. and A. Ahmad. 2023. “Saving Energy and Maintaining Indoor Comfort Level: A Reinforcement Learning Based Approach”. In 2023 IEEE 11th International Conference on Smart Energy Grid Engineering (SEGE), 240–247, Oshawa (Canada): IEEE. doi: 10.1109/SEGE59172.2023.10274587. 36 A. ALONGI ET AL. Budaiwi, I.M., and M.A. Fasi. 2023. “Assessing the Energy-saving Potential and visual Comfort of Electrochromic Smart Windows in Office Buildings: A case study in Dhahran, Saudi Arabia.” Sustainability 15: 9632. doi: 10.3390/ su15129632. Calì, D., P. Matthes, K. Huchtemann, R. Streblow, and D. Müller. 2015. “CO 2 Based Occupancy Detection Algorithm: Experimental Analysis and Validation for Office and Residential Buildings.” Building and Environment 86: 39–49. doi: 10.1016/j.buildenv.2014.12.011. Calvo, I., A. Espin, J.M. Gil-García, P.F. Bustamante, O. Barambones, and E. Apiñaniz. 2022. “Scalable IoT architecture for Monitoring IEQ Conditions in Public and Private Buildings.” Energies 15: 2270. doi: 10.3390/ en15062270. Candanedo, L.M., and V. Feldheim. 2016. “Accurate Occupancy Detection of an Office Room from Light, Temperature, Humidity and CO 2 Measurements Using Statistical Learning Models.” Energy and Buildings 112: 28–39. doi: 10.1016/j.enbuild.2015.11.071. Capozzoli, A., M.S. Piscitelli, A. Gorrino, I. Ballarini, and V. Corrado. 2017. “Data Analytics for Occupancy Pattern Learning to Reduce the Energy Consumption of HVAC systems in Office Buildings.” Sustainable Cities and Society 35: 191–208. doi: 10.1016/j.scs.2017.07.016. Carlucci, S., F. Causone, F.D. Rosa, and L. Pagliano. 2015. “A Review of Indices for Assessing Visual Comfort with a View to their Use in Optimization Processes To Support Building Integrated Design.” Renewable and Sustainable Energy Reviews 47: 1016–1033. doi: 10.1016/J.RSER.2015.03.062. Cascone, Y., M. Ferrara, L. Giovannini, and G. Serale. 2017. “Ethical Issues of Monitoring Sensor Networks for Energy Efficiency in Smart Buildings: A Case Study.” Energy Procedia 134: 337–345. doi: 10.1016/j.egypro.2017.09.540. Chen, D., S. Barker, A. Subbaswamy, D. Irwin, and P. Shenoy. 2013. “Non-Intrusive Occupancy Monitoring Using Smart Meters.” in: Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings, ACM, 1–8. doi: 10.1145/2528282.2528294. Chen, L., Q. Xu, Y. Yang, and J. Song. 2021. “Optimal Energy Management of Smart Building for Peak Shaving Considering Multi-energy Flexibility Measures.” Energy and Buildings 241: 110932. doi: 10.1016/ j.enbuild.2021.110932. Chen, X., Q. Wang, and J. Srebric. 2016. “Occupant Feedback Based Model Predictive Control for Thermal Comfort and Energy Optimization: A Chamber Experimental Evaluation.” Applied Energy 164: 341–351. doi: 10.1016/ j.apenergy.2015.11.065. Chen, Z., J. Xu, and Y.C. Soh. 2015. “Modeling Regular Occupancy in Commercial Buildings Using Stochastic Models.” Energy and Buildings 103: 216–223. doi: 10.1016/j.enbuild.2015.06.009. Chen, Z., R. Zhao, Q. Zhu, M.K. Masood, Y.C. Soh, and K. Mao. 2017. “Building Occupancy Estimation with Environmental Sensors via CDBLSTM.” IEEE Transactions on Industrial Electronics 64: 9549–9559. doi: 10.1109/ TIE.2017.2711530. Cheng, C.-C., and D. Lee. 2014. “Smart Sensors Enable Smart Air Conditioning Control.” Sensors 14: 11179–11203. doi: 10.3390/s140611179. Choi, J.-H., and D. Yeom. 2017. “Study of Data-driven Thermal Sensation Prediction Model as a Function of Local Body Skin Temperatures in a Built Environment.” Building and Environment 121: 130–147. doi: 10.1016/ j.buildenv.2017.05.004. Choi, J.-H., and R. Zhu. 2015. “Investigation of the Potential Use of Human Eye Pupil Sizes to Estimate Visual Sensations in the Workplace Environment.” Building and Environment 88: 73–81. doi: 10.1016/ j.buildenv.2014.11.025. Choi, J.-H., V. Loftness, and A. Aziz. 2012. “Post-Occupancy Evaluation of 20 Office Buildings as basis for future IEQ Standards and Guidelines.” Energy and Buildings 46: 167–175. doi: 10.1016/j.enbuild.2011.08.009. Clements, N., R. Zhang, A. Jamrozik, C. Campanella, and B. Bauer. 2019. “The Spatial and Temporal Variability of the Indoor Environmental Quality During Three Simulated Office Studies at A Living Lab.” Buildings 9: 62. doi: 10.3390/buildings9030062. Dahlan, A.S., and M.A. Eissa. 2015. “The Impact of day Lighting in Classrooms on Students' Performance.” International Journal of Soft Computing and Engineering ((IJSCE)) 4: 2231–2307. Datasheet, K 30 Senseair – An Asahi Kasei Group Company. 2022. https:// rmtplusstoragesenseair.blob.core.windows.net/docs/publicerat/PSP110.pdf. Datasheet, BME680. 2024 Bosch Sensortec. https://www.bosch-sensortec.com/products/environmental-sensors/gassensors/bme680/. Datasheet, CCS811. 2020. Joy-IT. https://joy-it.net/files/files/Produkte/SEN-CCS811V1/SEN-CCS811V1-Datasheet02.11.2020.pdf. Datasheet, DHT22. 2012. Aosong (Guangzhou) Electronics Co., Ltd. https://cdn-shop.adafruit.com/datasheets/ DHT22.pdf. Datasheet, GP2Y1010AU0F. 2018. SHARP/Socle Technology.SHARP/Socle Technology. Datasheet, Grove - sound sensor.2015. Seeed Studio. https://wiki.seeedstudio.com/Grove-Sound_Sensor/. Datasheet, S8, Senseair – An Asahi Kasei Group Company. 2022. https:// rmtplusstoragesenseair.blob.core.windows.net/docs/publicerat/PSP107.pdf. INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 37 Datasheet, Telaire T9602. 2018. Amphenol Advanced Sensors. https://www.mouser.com/datasheet/2/18/AAS-920-63 8H-Telaire-T9602-041318-web-1315842.pdf?srsltid= AfmBOooBJ-QgiYngnzsJevK29pCe-gI3laSylPlA9Aa0gY5 EbVvJQdpV. Datasheet, TGS2602. 2018. Figaro Engineering Inc. Figaro Engineering Inc. Datasheet, TSL2561. 2025. Adafruit Industries LCC. https://learn.adafruit.com/tsl2561. Daum, D., F. Haldi, and N. Morel. 2011. “A Personalized Measure of Thermal Comfort for Building Controls.” Building and Environment 46: 3–11. doi: 10.1016/j.buildenv.2010.06.011. Davis, J.A., and D.W. Nutter. 2010. “Occupancy Diversity Factors for Common University Building Types.” Energy and Buildings 42: 1543–1551. doi: 10.1016/j.enbuild.2010.03.025. Dawe, M., P. Raftery, J. Woolley, S. Schiavon, and F. Bauman. 2020. “Comparison of Mean Radiant And Air Temperatures in Mechanically-conditioned Commercial Buildings from over 200,000 field and Laboratory Measurements.” Energy and Buildings 206: 109582. doi: 10.1016/j.enbuild.2019.109582. Deng, Z., B. Dong, X. Guo, and J. Zhang. 2024. “Impact of Indoor Air Quality and Multi-domain Factors on Human Productivity and Physiological Responses: A Comprehensive Review.” Indoor Air 2024: 1–25. doi: 10.1155/2024/ 5584960. Derbez, M., B. Berthineau, V. Cochet, M. Lethrosne, C. Pignon, J. Riberon, and S. Kirchner. 2014. “Indoor Air Quality and Comfort in Seven Newly Built, Energy-efficient Houses in France.” Building and Environment 72: 173–187. doi: 10.1016/j.buildenv.2013.10.017. Derbez, M., B. Berthineau, V. Cochet, C. Pignon, J. Ribéron, G. Wyart, C. Mandin, and S. Kirchner. 2014. “A 3-Year Follow-up of Indoor Air Quality and Comfort in Two Energy-efficient Houses.” Building and Environment 82: 288–299. doi: 10.1016/j.buildenv.2014.08.028. Devitofrancesco, A., L. Belussi, I. Meroni, and F. Scamoni. 2019. “Development of an Indoor Environmental Quality Assessment Tool for the Rating of Offices in Real Working Conditions.” Sustainability 11: 1645. doi: 10.3390/ su11061645. Diraco, G., A. Leone, and P. Siciliano. 2015. “People Occupancy Detection and Profiling with 3D Depth Sensors for Building Energy Management.” Energy and Buildings 92: 246–266. doi: 10.1016/j.enbuild.2015.01.043. D'Oca, S., and T. Hong. 2015. “Occupancy Schedules Learning Process Through a Data Mining Framework.” Energy and Buildings 88: 395–408. doi: 10.1016/j.enbuild.2014.11.065. Dong, B., and B. Andrews. 2009. “Sensor-Based Occupancy Behavioral Pattern Recognition for Energy and Comfort Management in Intelligent Buildings.” In Proceedings of Building Simulation 2009: 11th Conference of IBPSA,1444–1451. IBPSA. doi: 10.26868/25222708.2009.1444-1451. Dong, B., and K.P. Lam. 2011. “Building Energy and Comfort Management Through Occupant Behaviour Pattern Detection Based on a Large-scale Environmental Sensor Network.” Journal of Building Performance Simulation 4: 359–369. doi: 10.1080/19401493.2011.577810. Dong, B., B. Andrews, K.P. Lam, M. Höynck, R. Zhang, Y.-S. Chiou, and D. Benitez. 2010. “An Information Technology Enabled Sustainability Test-bed (ITEST) for Occupancy Detection Through an Environmental Sensing Network.” Energy and Buildings 42: 1038–1046. doi: 10.1016/j.enbuild.2010.01.016. Dong, B., V. Prakash, F. Feng, and Z. O'Neill. 2019. “A review of Smart Building Sensing System for Better Indoor Environment Control.” Energy and Buildings 199: 29–46. doi: 10.1016/j.enbuild.2019.06.025. Duarte, C., K.V.D. Wymelenberg, and C. Rieger. 2013. “Revealing Occupancy Patterns in an office Building Through the use of Occupancy Sensor Data.” Energy and Buildings 67: 587–595. doi: 10.1016/j.enbuild.2013.08.062. Ekwevugbe, T., N. Brown, V. Pakka, and D. Fan. 2017. “Improved Occupancy Monitoring in Non-Domestic Buildings.” Sustainable Cities and Society 30: 97–107. doi: 10.1016/j.scs.2017.01.003. Thorseth, A.2024. “EN 12665:2024. Light and Lighting – Basic Terms and Criteria for Specifying Lighting Requirements. European Committee for Standardization. https://orbit.dtu.dk/en/publications/en-126652024light-and-lighting-basic-terms-and-criteria-for-spec. Erickson, V.L., and A.E. Cerpa. 2010. “Occupancy Based Demand Response HVAC Control Strategy.” In Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building ACM, 7–12. 10.1145/ 1878431.1878434. Erickson, V.L., and A.E. Cerpa. 2012. “Thermovote: Participatory Sensing for Efficient Building HVAC Conditioning.” In Proceedings of the Fourth ACM Workshop on Embedded Sensing Systems for EnergyEfficiency in Buildings, 9–16. doi: 10.1145/2422531.2422534. European Commission, 2018. “Directive (EU) 2018/844 of the European Parliament and of the Council of 30 May 2018 amending Directive 2010/31/EU on the Energy Performance of Buildings and Directive 2012/27/EU on Energy Efficiency.” Official Journal of the European Union L 156: 75–95. https://eur-lex.europa.eu/eli/dir/2018/ 844/oj/eng. European Commission, Directive (EU). 2024. "2024/1275 on the energy performance of buildings (recast)." Official Journal of the European Union. European Commission, Smart Readiness IndicatorAccessed: 2025-07-26. https://energy.ec.europa.eu/topics/energyefficiency/energy-performance-buildings/smart-readiness-indicator_en. Fanger, P. 1967. “Calculation of Thermal Comfort: Introduction of a Basic Comfort Equation.” ASHRAE Transactions 73. III4.1–III4.20. 38 A. ALONGI ET AL. Feng, F., and Z. Li. 2017. “A Methodology to Identify Multiple Equipment Coordinated Control with Power Metering System Energy Procedia .” In 8th International Conference on Applied Energy, ICAE2016, 8-11 October 2016, Beijing, China, 105: 2499–2505. doi: 10.1016/j.egypro.2017.03.721. Feyzi, M., and H. Mojallali. 2024. “Optimal Placement of Light Sensor for Improving Energy Efficiency and Visual Comfort in Smart Buildings.” e-Prime – Advances in Electrical Engineering, Electronics and Energy 9: 100681. doi: 10.1016/j.prime.2024.100681. Fisk, W.J. 2000. “Health and Productivity Gains from Better Indoor Environments and Their Relationship with Building Energy Efficiency.” Annual Review of Energy and the Environment 25: 537–566. doi: 10.1146/ annurev.energy.25.1.537. Floris, A., S. Porcu, R. Girau, and L. Atzori. 2021. “An IoT-Based Smart Building Solution for Indoor Environment Management and Occupants Prediction.” Energies 14: 2959. doi: 10.3390/en14102959. Focchi S.r.l. – Elisya[online]. Accessed 17/03/2025. https://www.focchi.it/ww/elisya. Frontczak, M., and P. Wargocki. 2011. “Literature Survey on How Different Factors Influence Human Comfort in Indoor Environments.” Building and Environment 46: 922–937. doi: 10.1016/j.buildenv.2010.10.021. Fybra S.R.L. – Fybra Home [online]. Accessed on 17/03/2025. URL https://fybra.co/en/fybra-home/. Galcık, F. and R. Gargalik. 2013. “Real-Time Depth Map Based People Counting.” In 15th International Conference ACIVS, edited by J. Blanc-Talon, A. Kasinski, W. Philips, D. Popescu and P. Scheunders, Vol. 8192, 330–341. Springer International Publishing. doi: 10.1007/978-3-319-02895-8. Ganesh, G.A., S.L. Sinha, T.N. Verma, and S.K. Dewangan. 2021. “Investigation of Indoor Environment Quality and Factors Affecting Human Comfort: A Critical Review.” Building and Environment 204: 108146. doi: 10.1016/ j.buildenv.2021.108146. Geng, Y., W. Ji, Z. Wang, B. Lin, and Y. Zhu. 2019. “A Review of Operating Performance In Green Buildings: Energy Use, Indoor Environmental Quality and Occupant Satisfaction.” Energy and Buildings 183: 500–514. doi: 10.1016/ j.enbuild.2018.11.017. Genkin, M., and J. McArthur. 2023. “B-smart: A Reference Architecture for Artificially Intelligent Autonomic Smart Buildings.” Engineering Applications of Artificial Intelligence 121: 106063. doi: 10.1016/j.engappai.2023.106063. Ghaddar, N., K. Ghali, and S. Chehaitly. 2011. “Assessing Thermal Comfort of Active People in Transitional Spaces in Presence of Air Movement.” Energy and Buildings 43: 2832–2842. doi: 10.1016/j.enbuild.2011.06.040. Ghahramani, A., F. Jazizadeh, and B. Becerik-Gerber. 2014. “A Knowledge Based Approach for Selecting Energyaware and Comfort-driven Hvac Temperature Set Points.” Energy and Buildings 85: 536–548. doi: 10.1016/ j.enbuild.2014.09.055. Ghahramani, A., C. Tang, and B. Becerik-Gerber. 2015. “An Online Learning Approach or Quantifying Personalized Thermal Comfort Via Adaptive Stochastic Modeling.” Building and Environment 92: 86–96. doi: 10.1016/ j.buildenv.2015.04.017. Goyal, S., P. Barooah, and T. Middelkoop. 2015. “Experimental Study of Occupancy-based Control of HVAC zones.” Applied Energy 140: 75–84. doi: 10.1016/j.apenergy.2014.11.064. Gruber, M., A. Trüschel, and J.-O. Dalenbäck. 2014. “CO 2 Sensors for Occupancy Estimations: Potential in Building Automation Applications.” Energy and Buildings 84: 548–556. doi: 10.1016/j.enbuild.2014.09.002. Gu, T., Z. Fang, Z. Yang, P. Hu, and P. Mohapatra. 2019. “MMSense: Multi-Person Detection and Identification via mmwave Sensing.” In Proceedings of the 3rd ACM Workshop on Millimeter-wave Networks and Sensing Systems, ACM, 45–50. doi: 10.1145/3349624.3356765. Guo, X., K. Lee, Z. Wang, and S. Liu. 2021. “Occupants’ satisfaction with LEEDand non-LEED-certified apartments using social media data.” Building and Environment 206: 108288. doi: 10.1016/j.buildenv.2021.108288. Guo, X., D. Tiller, G. Henze, and C. Waters. 2010. “The Performance of Occupancy-based Lighting Control Systems: A Review.” Lighting Research Technology 42: 415–431. doi: 10.1177/1477153510376225. Habiba, U., I. Ahmed, M. Asif, H.H. Alhelou, and M. Khalid. 2024. “A Review on Enhancing Energy Efficiency and Adaptability Through System Integration for Smart Buildings.” Journal of Building Engineering 89: 109354. doi: 10.1016/j.jobe.2024.109354. Habibi, S.. 2020. “A Review of ICT-Based Applications to Monitor Environmental Parameters.” Construction Innovation 20: 112–127. doi: 10.1108/CI-05-2019-0043. Haipeng, L., Z. Xingyue, Z. Anfu, L. Liang, M. Huadong, H.P. Liu, X.Y. Zhang, and A.F. Zhou. 2021. “Indoor Environment and Indoor Environment and human sensing via Millimeter Human Sensing via Millimeter Wave Radio: A review.” ZTE Communications 19: 22–29. doi: 10.12142/ZTECOM.202103004. Heidary, R., J.P. Rao, and O.J.P. Fischer. 2023. “Smart Buildings in the IoT era: Necessity, Challenges, and Opportunities.” In Handbook of Smart Energy Systems, 1915–1935. Cham: Springer International Publishing. doi: 10.1007/978-3-030-97940-9_115. Heinzerling, D., S. Schiavon, T. Webster, and E. Arens. 2013. “Indoor Environmental Quality Assessment Models: A Literature Review and A Proposed Weighting and Classification Scheme.” Building and Environment 70: 210–222. doi: 10.1016/J.BUILDENV.2013.08.027. Hirt, J., T. Nordhausen, T. Fuerst, H. Ewald, and C. Appenzeller-Herzog. 2024. “Guidance on Terminology, Application, and Reporting of Citation Searching: The Tarcis Statement.” BMJ. e078384. doi: 10.1136/bmj-2023078384. INTERNATIONAL JOURNAL OF SUSTAINABLE ENERGY 39 Hnat, T.W., E. Griffiths, R. Dawson, and K. Whitehouse. 2012. “Doorjamb: Unobtrusive Room-level Tracking of People in Homes Using Doorway Sensors.” In Proceedings of the 10th ACM Conference on Embedded Network Sensor Systems, ACM, 309–322. doi: 10.1145/2426656.2426687. Hongisto, V. 2005. “A Model Predicting the Effect of Speech of Varying Intelligibility on Work Performance.” Indoor Air 15: 458–468. doi: 10.1111/j.1600-0668.2005.00391.x. Horr, Y.A., M. Arif, M. Katafygiotou, A. Mazroei, A. Kaushik, and E. Elsarrag. 2016. “Impact of Indoor Environmental Quality on Occupant Well-being1269 and comfort: A Review of the Literature.” International Journal of Sustainable Built Environment 5: 1–11. doi: 10.1016/j.ijsbe.1270 2016.03.006. Huang, Y., and M. J. Griffin. 2012. “The Effects of Sound Level and Vibration Magnitude on the Relative Discomfort of Noise and Vibration.” The Journal of the Acoustical Society of America 131: 4558–4569. doi: 10.1121/1.4705285. Jia, R., R. Dong, S.S. Sastry, and C.J. Spanos. 2017. “Privacy-Enhanced Architecture for Occupancy-based HVAC Control.” In Proceedings of the 8th International Conference on Cyber-Physical Systems ACM, 177–186. doi: 10.1145/3055004.3055007. Jiang, C., M.K. Masood, Y.C. Soh, and H. Li. 2016. “Indoor Occupancy Estimation from Carbon Dioxide Concentration.” Energy and Buildings 131: 132–141. doi: 10.1016/j.enbuild.2016.09.002. Jin, M., S. Liu, S. Schiavon, and C. Spanos. 2018. “Automated Mobile Sensing: Towards high-Granularity Agile Indoor Environmental Quality Monitoring.” Building and Environment 127: 268–276. doi: 10.1016/ j.buildenv.2017.11.003. Jung, W., and F. Jazizadeh. 2019. “Human-in-the-loop HVAC operations: A Quantitative Review on Occupancy, Comfort, and Energy-efficiency Dimensions.” Applied Energy 239: 1471–1508. doi: 10.1016/j.apenergy.2019.01.070. Karimi, H., M.A. Adibhesami, H. Bazazzadeh, and S. Movafagh. 2023. “Green buildings: Human-Centered and Energy Efficiency Optimization Strategies.” Energies 16: 3681. doi: 10.3390/en16093681. Kauneliené, V., T. Prasauskas, E. Krugly, I. Stasiulaitienė, D. Čiužas, L. Šeduikytė, and D. Martuzevičius. 2016. “Indoor Air Quality in Low Energy Residential Buildings in Lithuania.” Building and Environment 108: 63–72. doi: 10.1016/j.buildenv.2016.08.018. Kaushik, A., M. Arif, P. Tumula, and O.J. Ebohon. 2020. “Effect of Thermal Comfort on Occupant Productivity in Office Buildings: Response Surface Analysis.” Building and Environment 180: 107021. doi: 10.1016/ j.buildenv.2020.107021. Kenny, T. 2005. “Chapter 1 - Sensor Fundamentals.” In Wilson J., Sensor Technology Handbook, pp. 1–20. Newnes. doi: 10.1016/B978-075067729-5/50041-0. Khovalyg, D., O.B. Kazanci, H. Halvorsen, I. Gundlach, W.P. Bahnfleth, J. Toftum, and B.W. Olesen. 2020. “Critical Review of Standards for Indoor Thermal Environment and Air Quality.” Energy and Buildings 213: 109819. doi: 10.1016/J.ENBUILD.2020.109819. Kim, H., and T. Hong. 2024. “Emotion-Oriented Recommender System for Personalized Control of Indoor Environmental Quality.” Building and Environment 254: 111396. doi: 10.1016/j.buildenv.2024.111396. Kim, Y.-S., and J. Srebric. 2017. “Impact of Occupancy Rates on the Building Electricity Consumption in Commercial Buildings.” Energy and Buildings 138: 591–600. doi: 10.1016/j.enbuild.2016.12.056. Kim, J., S. Schiavon, and G. Brager. 2018a. “Personal Comfort Models–A New Paradigm in Thermal Comfort for Occupant-centric Environmental Control.” Building and Environment 132: 114–124. doi: 10.1016/ j.buildenv.2018.01.023. Kim, J., Y. Zhou, S. Schiavon, P. Raftery, and G. Brager. 2018b. “Personal Comfort Models: Predicting Individuals’ Thermal Preference Using Occupant Heating and Cooling Behavior and Machine Learning.” Building and Environment 129: 96–106. 10.1016/j.buildenv.2017.12.011. Kleiminger, W., F. Mattern, and S. Santini. 2014. “Predicting household occupancy for smart heating control: A comparative performance analysis of state-of-the-art approaches.” Energy and Buildings 85: 493–505. doi: 10.1016/ j.enbuild.2014.09.046. Ko, W.H., S. Schiavon, S. Altomonte, M. Andersen, A. Batool, W. Browning, G. Burrell, et al. 2022. “Window View Quality: Why it Matters and What We Should Do.” LEUKOS – Journal of Illuminating Engineering Society of North America 18: 259–267. doi: 10.1080/15502724.2022.2055428. Kofler, M.J., C. Reinisch, and W. Kastner. 2012. “A Semantic Representation of Energy-related Information in Future Smart Homes.” Energy and Buildings 47: 169–179. doi: 10.1016/j.enbuild.2011.11.044. Kong, M., J. An, D. Jung, and T. Hong. 2023. “Occupant-Centered Indoor Environmental Quality Management: Physiological Response Measuring Methods.” Building and Environment 243: 110661. doi: 10.1016/ j.buildenv.2023.110661. Kontes, G., G. Giannakis, P. Horn, S. Steiger, and D. Rovas. 2017. “Using Thermostats for Indoor Climate Control in Office Buildings: The Effect on Thermal Comfort.” Energies 10: 1368. doi: 10.3390/en10091368. Kumar, A., S. Sharma, N. Goyal, A. Singh, X. Cheng, and P. Singh. 2021. “Secure and Energy-efficient Smart Building Architecture with Emerging Technology IoT.” Computer Communications 176: 207–217. doi: 10.1016/ j.comcom.2021.06.003. Kumar, P., C. Martani, L. Morawska, L. Norford, R. Choudhary, M. Bell, and M. Leach. 2016. “Indoor Air Quality and Energy Management Through Real-time Sensing in Commercial Buildings.” Energy and Buildings 111: 145–153. doi: 10.1016/j.enbuild.2015.11.037. 40 A. ALONGI ET AL.