Conceptual Design of IT Solution for Multimodality and Traffic Flow Optimization in Catchment Area of the Airport: Case Study of Zagreb Airport

Conceptual Design of IT Solution for Multimodality and Traffic Flow Optimization in Catchment Area of the Airport: Case Study of Zagreb Airport Saša Vojvodić*, Krešimir Vidović*, Domagoj Leljak*, Ana Najev* * Ericsson Nikola Tesla, Zagreb, Croatia [email protected] Abstract — Multimodality and traffic flow optimization IT solution (platform) is a software-based decision support system intended to be used by relevant stakeholders responsible for strategic and comprehensive mobility management. The platform provides environment for design and an implementation of transport policies that tackle urban mobility problems by encouraging a shift towards more sustainable transport modes, which is in line with generally accepted transport related environmental, social, and economic objectives. This paper will propose conceptual design of IT solution that will be applied in the catchment area of Zagreb Airport. The proposed architecture includes a three-layer approach (data sources, data aggregation, application). Following the definition of mobility management analytical use cases identified and set by relevant stakeholders, this paper will define requirements and propose solution for distinct layers (primary data sources and data alternatives, data aggregation layer and application layer requirements) together with definition of integration and validation points within the system and process. Since the effect of the proposed and implemented transport policies should be measurable, this paper will also provide a definition of related key performance indicators as well as the methodology of their calculations within the defined processes. Keywords - traffic flow optimization, IT solution for mobility management I. INTRODUCTION European Union's Sustainable and Smart Mobility Strategy states that the unobstructed movement of people and goods across its internal borders is a fundamental principle. In light of this, pledge to significantly reduce emissions and increase sustainability has been the biggest challenge the transport sector has been facing recently. [1] The strategy outlined by the EU aims at applying smart digital solutions like the intelligent transport systems (ITS) in hopes of transforming the transport sector into a truly multimodal system of sustainable and smart mobility services. A number of research papers have covered this topic and solutions to current challenges are actively investigated. [2]–[4] In urban and sub-urban areas supporting zero-emission airports these multimodal digital solutions are vital for a successful strategical implementation of solutions. [1] This paper outlines the conceptual design of an IT solution which will be applied on the catchment area of Zagreb airport. The catchment area of an airport is the surrounding area from which passengers arrive [5], along with other airport user types which are also taken into consideration, like employees, greeters, buyers, etc. Mobility stakeholders have a challenging task to objectively assess the effects that implementation of certain policies and technologies can have. Moreover, it can be difficult to apply the same solutions to different and disparate urban areas. [6] Therefore, in order to create more impactful policies, a section of the IT solution is dedicated to calculations of key performance indicators (KPI’s). This provides stakeholders with a supporting tool for a more transparent and mindful process of creating and implementing transport policies, along with the ability to continuously assess and monitor defined KPI’s. II. STATE OF THE ART A detailed conceptualization and proposed realization of an intelligent transport system for urban mobility management has been extensively described in three recent papers [7]–[8]. The work focuses on solutions for small to medium sized cities. Authors represent the architecture of their data-driven solution for urban traffic management as consisting of three layers. The first layer comprises of the required data sources and is discussed in the paper by D. Medved et al. [8]. Their report focuses on necessary data needed for a well-organized decision-making model. The input data includes conventional traffic, ecological and meteorological data, along with static and dynamic public transport data supported by additional traffic light and traffic flow information. The second layer in the architecture consists of a data aggregation platform which is at the heart of the software-based decision-making system [9]. The platform enables large amounts of data to be collected from different sources and generated by various stakeholders, which can be further used with high interoperability. Additionally, in this paper authors give a detailed description of main stakeholders and their 1030 MIPRO 2024, May 20 - 24, 2024, Opatija, Croatia 2024 47th MIPRO ICT and Electronics Convention (MIPRO) | 979-8-3503-8250-1/24/$31.00 ©2024 IEEE | DOI: 10.1109/MIPRO60963.2024.10569931 Authorized licensed use limited to: Universitatea Tehnica din Cluj-Napoca. Downloaded on December 16,2025 at 08:14:17 UTC from IEEE Xplore. Restrictions apply. methodology based on the Delphi method for defining requirements and process prioritization. Their platform is highly customized to fulfil different requirements: defining data standards and protocols, storage, management and permissions, and creating supported transport protocols for data transport. At the top layer of the urban mobility management system’s architecture, one finds various applications which concurrently cover a plethora of information needed to propose solutions for traffic management. The concept for developing prototypes of different application modules is reported on in detail by M. Draganic et al. [7]. Authors construe the identified requirements for the software application and systemize and prioritize user aspirations and needs based on stakeholder input. The proposed technical solution includes functional and component architecture following the principles of software building. III. SYSTEM ARCHITECTURE Since the qualitative properties of datasets appropriated for the case study of Zagreb airport were extrapolated on in a previous publication [10], here we briefly summarize the data categories and their respective features: - Traffic data: o Exchanged in DATEX II format; o Obtained in a business-to-business (b2b) setup; o Contains real-time localized traffic variables such as vehicle flow rate, average velocity and headway distance. - Public transport data: o Exchanged in NeTEx/GTFS format; o Publicly available/b2b; o Contains information on transport lines, schedules, and stops. - Anonymized mobile network data: o A synthesis of multiple mobile network data sources: Charge/Call Data Records (CDR), Cell Trace Records (CTR), subscription database, and Performance/Configuration Management (PM/CM); o Heavily scrutinized under EU GDPR laws on multiple levels, and as such only available for limited periods of time, and fully anonymized; o Comes in form of network snapshots usually temporally separated by at least a month, and lasting a maximum of one week; o Contains information on aggregated user mobility. - Airport flight data: o Contains information on airline flight variables – arrival/departure schedules, flight length, carrier data and more; o Static information. - Airport parking data: o Real time information on the airport parking status – entries and exits, user type, etc. - Other data: o Methodology validation data, census, statistical estimates… Previously, our data aggregation platform has also been described [9], so we only give an outlining rundown here, with a further description of a layer below, specifically a layer designed as a data preprocessing platform. We focus here only on the mobile network dataflow, since it anyway comprises the largest and most complex dataset category in this case study. The dataflow in the processing layer is as follows: first, unfiltered fully anonymized mobile network data is ingested through an SFTP protocol and processed in a containerized Apache Spark/Hadoop environment. The Apache Spark has proven to be the most stable and efficient engine for processing tasks of requisite size (order of magnitude ~up to few TB’s per network snapshot) and complexity. We use the PySpark Python API when interfacing with Spark, both due to ease of use, and ease of integration. After the processing, data is stored in an easily accessible Hadoop Distributed File System in parquets. Parameters describing the allocation of resources are set prior to each data processing task, due to the varying complexity of each mobile network snapshot (e.g., Croatia is visited by 10-15 million tourists annually, most of them visiting during summer, so the network load varies wildly through the year). All the physical hardware (disks, processors…) is kept locally, i.e., local physical infrastructure is employed. Out of this very large initial dataset, a few parquets processed using different filters are considered for true output, each intended for a different use case. A reduced (use case specific) dataset is engineered out of the aforementioned true output, and using the HTTPS protocol, ingested into our three-layer data aggregation platform in the JSON format. The data aggregation platform is deployed in an Oracle Cloud environment, and holds most of the data in Apache Cassandra/Spark databases, with accommodating metadata in the form of preprovisioned hierarchies stored in a PostgreSQL database. The northmost layer is used to communicate with external applications, some of which were designed as a support for some aspects of this case study (e.g., a GIS migration-tracking web application serving as an analytical tool intended for traffic-optimization planning). The output interface layer is intentionally kept as modular as possible due to the very wide spectrum of use cases this synthesized dataset potentially facilitates. Since some of the data sets employed in the case study might be considered sensitive, all organisational measures, technical procedures and guiding principles for building peoples trust in usage of such data were taken into account while processing this data. [11] IV. KPI MEASUREMENT FRAMEWORK Within the H2020 project OLGA (hOlistic & Green Airports, grant number 101036871) an IT solution for multimodal traffic optimization by means of data fusion from three different sources has been proposed. An analytical use case has been identified with the aim of 1031 Authorized licensed use limited to: Universitatea Tehnica din Cluj-Napoca. Downloaded on December 16,2025 at 08:14:17 UTC from IEEE Xplore. Restrictions apply. optimizing current or introducing new public transport lines leading up to the gravitational area of Zagreb airport (Velika Gorica) which would benefit residents, passengers and employees alike. [10] Alongside this use case, key performance indicators were identified to help assess the performance of measures taken in the OLGA project. A key KPI addressing the general environmental performance of the OLGA project is based on determining the effect an extension of the multimodal transport network can have on users. A larger and ramified network area helps more users reach the desired destination by different means and routes and can additionally reduce their travel time. Two key indicators are established: the extension of the network (1) and its maximum capacity (2). The linear length is a reliable measure for the size of a transport network since the area it insists on can be limited by geographical factors or the size of the considered airport. Linear length can be easily obtained through direct means or simply using tools available online. The maximum capacity of the network depends on the concurrent active vehicle number and can be retrieved from information given by network operators or through direct observation. [12] MS1 .1 : Ext= Length [km] (1) MS1.2 : Capmax=Nvehiclesmax*Capvehicle [pax/h] (2) Our collaboration on the research project requests KPI evaluation for which we analyse points of interconnection between processes and possible realization of identified use cases along with creating required outputs. Methodology proposed for the realization of public transport optimization based on dedicated stakeholder input and the calculations for the corresponding KPI’s are outlined in Figure 1. The process starts by first analysing the existing public transport supply, based on the following data sources: public transport stops, lines and timetables from public transport operators (in standardized format like GTFS or NeTEx) and the data on vehicle characteristics (type, capacity) needed for KPI calculation. Following the methodology, we are able to identify spatial areas (sectors, locations) covered by existing public transport lines serving the airport and calculate the network length and transport capacity. Additionally, by using anonymized telecom data sets which contain the origin destination matrix and the user type data an analysis of the transport demand can be performed. Spatial information is obtained with the origin destination matrix which gives us transitions between sectors and determines relevant routes. This enables us to identify frequent sectors of users’ origins whose destinations are in the airport sector and the transport modes in between them. Analysis of Public Transport Offer PT Data Stops Lines Timetables Public transport optimization for residents/airport users Analysis of Transport Demand Telco Data Origin Destinatio n Matrix User Type Proposal of measures Output Final Proposal of new PT lines Final proposal of Modification of existing PT lines Final proposal of Introduction of new stops Final proposal of Modifications to timetables Use Cases Analysis Identifications of spatial sectors (locations) covered by PT lines addressing airport Spatial data Sectors Identification of most frequent origins sectors ending at airport sector per user type and vice versa Spatial analysis (identification) of frequent origin/ destination sectors not covered by PT (delta) Temporal analysis of existing PT lines offer/ demand compliance Use case 1,2 ReportKPI MS1.1 MS1.2 (Existing) Public Transport network lenght New Public Transport Network Length New Public Transport Maximum Capacity (Existing) Public Transport Capacity O/D clustering per user type & in total Calculation of distance from cluster center to existing (targeted): 1) PT lines 2) Stations Identification of transport demand peeks Definition of criteria for PT network improvement Comparision of transport offer (timetables) Identification of gaps Vehicle data OSM road data Figure 1. Proposed methodology for analytical use case execution in relation with first KPI calculation Once the public transport supply has been determined and analysed, we are able to focus on the use case. By identifying frequent origin/destination sectors not covered by PT (delta) and conducting a temporal analysis of existing PT lines (offer/demand), compliance appropriate measures will be proposed. A goal is set to determine origin destination clustering per user type and in total. Furthermore, a detailed calculation of distances between origins and destination cluster centres is conducted by being mindful of the existing public transport lines and stations Based on the definition of criteria for public transport planning (expected walking distance from public transport stations, interconnection with other transport mode and infrastructure, etc...) a first set of measures will be identified along with a proposal for new PT lines in order to connect areas that are not currently connected. A modification of current locations of public transport stations and lines will be proposed with hopes of improving the current connectivity and infrastructure. We will offer a proposal for a modification of the timetables based on transport demand peeks and its comparison with the current offer. Following the validation by key stakeholders (city automotives, public transport operators) final measures will 1032 Authorized licensed use limited to: Universitatea Tehnica din Cluj-Napoca. Downloaded on December 16,2025 at 08:14:17 UTC from IEEE Xplore. Restrictions apply. be structured. Based on newly proposed measures, values required for KPI calculation will be identified, including new public transport length parameters and a new public transport maximum capacity. Second KPI (SOC3) is called “Traffic around airport”. It measures vehicles circulating around (entering/existing) the airport, on an hourly basis. Calculation of this KPI is based on equation (3). The methodology proposed by the project relies on the recorded traffic at the edges of the airport area, or on data collected from an array of cameras located along the main roads which approach the airport. The proposed assessment considers both public and private vehicles. SOC3 : NV[-/hour]; Δnv=(NVmeas-NVref)/NVref (3) Within the proposed concept of IT solution, following methodology is envisaged, as presented in Figure 2. Traffic data Vehicle count Traffic around airport - Vehicles circulating around (entering/ existing) the airport, on an hourly basis Use Cases Analysis Telco Data Origin Destinati on Matrix User Type Output Spatial data Sectors Identification of number of vehicles entering/exiting the airport Use case reportKPI SOC 3 OSM road data Virtual counters Airport data Parking data Vehicle speed data Identification of number of vehicles passing the airport Number of vehicles entering/exiting the airport parking Identification of trips performed by vehicle (no walking or cycling) )for all users with trips with airport sector as origin or destination of trip (optional) Number of vehicles passing the airport from traffic counter/virtual counter data Identification of trips performed by personal vehicles for all users with trips transiting airport sector (airport sector as visited sector or similar) Number of vehicles entering/exiting the airport (on hourly base) Number of vehicles passing the airport (on hourly base) Figure 2. Proposed methodology for analytical use case execution in relation with second KPI calculation For the KPI Traffic around airport - vehicles circulating around (entering/existing) the airport, on an hourly basis, we use the following data sources: Traffic data (vehicle count and vehicle speed data), airport data (parking data) and anonymised telecom data (origin destination matrix, user types, virtual traffic counters). Identified data sources will be used for a) identification of the number of vehicles entering/exiting the airport and b) identification of the number of vehicles passing the airport. For a) we identify trips performed by vehicles (no walking or cycling) for all users whose trips have the airport sector as the origin or destination (optional). This data is then associated with the number of vehicles entering/exiting the airport parking. For b), we identify trips performed by personal vehicles for all users which have visited the airport sector in transit (airport sector as visited sector or similar). This data is combined with the identified number of vehicles passing the airport detected by traffic counter/virtual counters. Finally, as the input for the calculation of the SOC3 KPI we use the (a) identified number of vehicles entering/exiting the airport on an hourly basis and (b) the number of vehicles passing the airport sector on an hourly basis. V. CONCLUSION This paper presents a conceptual design of an IT solution for a decision-making system which aims to help with proposing, implementing, and monitoring the effects of transport policies to be applied in the catchment area of Zagreb airport. The defined solution architecture consists of three layers: data sources, data processing & aggregation and the application. Such a transport flavour data aggregation platform enables utilisation of traffic data, public transport data, anonymised mobile network data and airport network data, and creates a base for data driven decision making using identified analytical use cases. Based on the example of optimisation of a public transport network by an increase of length of its network and capacity, a framework for evaluation of receptive KPI's has been proposed. Following research will be focused on the extension of analytical use cases, identification and integration of other data sources, and the extension of KPI’s framework. ACKNOWLEDGEMENT This paper presents the results of activities performed within the OLGA project. OLGA project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement № 101036871. REFERENCES [1] European Commision, “Sustainable and Smart Mobility Strategy – putting European transport on track for the future,” 2020. [2] Z. Mahrez, E. Sabir, E. Badidi, W. Saad, and M. Sadik, “Smart Urban Mobility: When Mobility Systems Meet Smart Data,” IEEE Trans. Intell. Transp. Syst., vol. 23, no. 7, pp. 6222–6239, Jul. 2022, doi: 10.1109/TITS.2021.3084907. [3] P. A. Maldonado Silveira Alonso Munhoz et al., “Smart Mobility: The Main Drivers for Increasing the Intelligence of Urban Mobility,” Sustainability, vol. 12, no. 24, p. 10675, Dec. 2020, doi: 10.3390/su122410675. [4] M. Šoštarić, K. Vidović, M. Jakovljević, and O. Lale, “Data-driven Methodology for Sustainable Urban Mobility Assessment and Improvement,” Sustain., 2021, [Online]. Available: https://www.bib.irb.hr/1138783 [5] R. Lieshout, “Measuring the size of an airport’s catchment area,” J. Transp. Geogr., vol. 25, pp. 27–34, Nov. 2012, doi: 10.1016/j.jtrangeo.2012.07.004. [6] I. Kaparias and M. G. H. Bell, “Key Performance Indicators for traffic management and Intelligent Transport Systems,” London, 2011. [7] M. Draganic et al., “Requirements on applications within the decision-making system for urban mobility management,” Transportation Research Procedia, vol. 64, no. C. pp. 60–67, 2022. doi: 10.1016/j.trpro.2022.09.008. [8] D. Medved, D. Blažinic, V. Galijan, and N. Antolovic, “Evolution of data sources for integrated data-driven urban mobility management,” in Transportation Research Procedia, 2022, pp. 68– 75. doi: 10.1016/j.trpro.2022.09.009. 1033 Authorized licensed use limited to: Universitatea Tehnica din Cluj-Napoca. Downloaded on December 16,2025 at 08:14:17 UTC from IEEE Xplore. Restrictions apply. [9] R. Radovic, I. Marasovic, V. Cackovic, D. Pleština, D. Keresteny, and Z. Anic, “The concept of a data aggregation platform in the function of a decision-making system for urban mobility management,” Transportation Research Procedia, vol. 64, no. C. pp. 53–59, 2022. doi: 10.1016/j.trpro.2022.09.007. [10] M. Tica, I. Štimac, K. Vidović, S. Vojvodić, and I. Stipanović, “Analytics Use Cases for Landside Traffic Optimization in the Catchment Area of the Airport: Case Study of Zagreb Airport,” in 2023 46th International Convention on Information, Communication and Electronic Technology (MIPRO), Opatija, Hrvatska, 2023. [Online]. Available: https://ieeexplore.ieee.org/abstract/document/10159821/ [11] R. Jansen et al., “Guiding principles to maintain public trust in the use of mobile operator data for policy purposes,” Data Policy, vol. 3, no. 3, 2021, doi: 10.1017/dap.2021.21. [12] Project OLGA, “WP1.1 OLGA KPI definition: master document,” 2022. 1034 Authorized licensed use limited to: Universitatea Tehnica din Cluj-Napoca. Downloaded on December 16,2025 at 08:14:17 UTC from IEEE Xplore. Restrictions apply.