Deliverable D5.2 Characteristics and classification of 6G device classes

A holistic flagship towards the 6G network platform and system, to inspire digital transformation, for the world to act together in meeting needs in society and ecosystems with novel 6G services. Deliverable D5.2 Characteristics and classification of 6G device classes Hexa-X-II project has received funding from the Smart Networks and Services Joint Undertaking (SNS JU) under the European Union’s Horizon Europe research and innovation programme under Grant Agreement No 101095759. Date of delivery: 31/10/2023 Version: 1.0 Project reference: 101095759 Call: HORIZON-JU-SNS-2022 Start date of project: 01/01/2023 Duration: 30 months Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 2 / 61 Document properties: Document Number: D5.2 Document Title: Characteristics and classification of 6G device classes Editor(s): Bipin Balakrishnan (EAB) Authors: Michiel Aernouts (IMEC), Pavlos Alexias (WIN), Azzam AlNahari (AAU), Bipin Balakrishnan (EAB), Madhawa Basnayaka (AAU), Panagiotis Demestichas (WIN), Claude Desset (IMEC), Jeroen Famaey (IMEC), Riku Jäntti (AAU), Alp Karakoc (AAU), Efstathios Katranaras (SEQ), Hamza Khan (LMF), Kalle Koskinen (AAU), Corentin Lavaud (QLC), Onel L. Alcaraz López (OUL), Harsha Master (NXP), Lukas W. Mayer (SAT), Nafiseh Mazloum (SON), Beatriz Mendes (UBW), Osmel Martínez Rosabal (OUL), Martin Schiefer (SAT), Mohamed Shehata (QLC), Hélio Simeão (UBW), Bikramjit Singh (LMF), Ritesh Kumar Singh (IMEC), Peerapol Tinnakornsrisuphap (QLC), Javier Velasquez (NXP), Christos Xygkos (WIN), Kun Zhao (SON). Contractual Date of Delivery: 31/10/2023 Dissemination level: PU1 Status: Draft Version: 1.0 File Name: Hexa-X-II_D5.2_final Revision History Revision Date Issued by Description V0.0 13.02.2023 Hexa-X-II WP5 Document creation, with initial ToC V0.1 23.08.2023 Hexa-X-II WP5 Draft, with contents based on D5.1 V0.2 11.09.2023 Hexa-X-II WP5 Draft, with internal WP review updates. V0.3 30.09.2023 Hexa-X-II WP5 Draft, with Cross-WP review updates V0.4 27.10.2023 Hexa-X-II WP5 Draft, with GA review updates. V1.0 30.10.2023 Hexa-X-II WP5 Final version ready for submission Abstract This report presents, based on Hexa-X-II use-cases defined so far, the characterization criteria, identified 6G device classes along with their characteristics and their technological enablers and provides an outlook for future considerations. 1 SEN = Sensitive, only members of the consortium (including the Commission Services). Limited under the conditions of the Grant Agreement PU = Public Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 3 / 61 Keywords 6G, Future devices, Device Classes, Device characteristics, Technology enablers, Sustainability considerations Disclaimer Funded by the European Union. The views and opinions expressed are however those of the author(s) only and do not necessarily reflect the views of Hexa-X-II Consortium nor those of the European Union or European Commission. Neither the European Union nor the granting authority can be held responsible for them. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 4 / 61 Executive Summary This report –” Characteristics and classification of 6G Device Classes” is the second deliverable from Work Package 5 (WP5). It analyzes the use-cases described in Hexa-X-II D1.1, complemented with similar ones described in Hexa-X project such as Hexa-X D1.2, Hexa-X D1.3 and Hexa-X D1.4, and defines the characterization criteria to identify 6G device classes and technological enablers to address WPO5.1. From a set of use-cases, this deliverable builds an understanding of key devices that realize a use-case while noting the deployment characteristics and its influence on the device operation. This helps to build a broad idea of the devices and how they will be integrated in a 6G System. This deliverable then defines a set of characterization criteria which serves as a basis to extract distinct characteristics that distinguish individual device classes from each other and other characteristics that these devices may share but to varying degrees. Furthermore, from the Hexa-X-II D1.1 deliverable, set of guidelines on the sustainable design is identified to serve as recommendations to these device classes. Four device classes are identified – energy neutral devices, reliable high data rate with bounded latency devices, highly reliable and low latency devices, and enhancements of massive Machine Type Communication (mMTC) devices. Energy neutrality will gain more prominence as the number of the devices continue to rapidly increase and with energy neutral device class, devices that rely only energy harvesting are addressed. They are recommended to generate zero e-waste, taking advantage of environment friendly material and manufacturing. Zero energy devices fall into this class. The Reliable high data rate with bounded latency device class caters to enabling immersive experiences and beyond what is possible in 5G. This device class covers the eXtended Reality (XR) devices and identifies the trade-off between availability and Quality of Experience (QoE). A third device class would be one with high reliability yet have increased data rate and less stringent latency than what is supported in 5G URLLC. These devices will have functional safety as a prominent non-communication criterion and cover the autonomous operating devices. Cobots, AGVs fall into this class. Lower data rates, yet with more reliability and availability than energy neutral devices lead to the fourth device class, which is seen as an enhancement of mMTC devices in 5G. They are battery operated, will enable low power and wide area IoT. From the current outlook, there could be additional devices that are evolutions of enhanced Mobile BroadBand (eMBB) and Ultra-Reliable and Low-Latency Communication (URLLC) devices. These devices emphasize the support for higher data rates & ultra-reliability respectively, potentially on new spectrum in centimetric range, sub-THz range and identifying the technology challenges they entail, and perhaps bringing in Machine Learning based techniques to find enhancements. Furthermore, there could be companion devices that support less (compute) capable devices by providing a secure, privacy aware compute connecting serving devices by flexible topologies. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 5 / 61 Table of Contents 1 Introduction ......................................................................................................................... 11 1.1 Scope and objective ..................................................................................................... 11 1.2 Methodology ................................................................................................................ 12 1.3 Structure of the document. ........................................................................................... 12 2 Motivating use-cases and services...................................................................................... 13 2.1 Use-cases ...................................................................................................................... 13 2.1.1 Autonomous supply chains ..................................................................................... 13 2.1.2 E-health for all ........................................................................................................ 14 2.1.3 Earth monitoring ..................................................................................................... 14 2.1.4 Immersive smart city ............................................................................................... 15 2.1.5 Fully merged cyber-physical worlds ....................................................................... 15 2.1.6 Interactive and cooperating mobile robots .............................................................. 16 2.1.7 Small coverage, low power micro-network in networks ........................................ 16 2.2 Services ........................................................................................................................ 16 2.2.1 Compute as a Service .............................................................................................. 16 2.2.2 Situation aware device reconfiguration ................................................................... 17 2.3 Summary of motivating use-cases and services analysis ............................................. 17 3 Characterization criteria and sustainability guidelines ................................................... 19 3.1 Characterization criteria ............................................................................................... 19 3.2 Sustainability guidelines .............................................................................................. 22 3.2.1 Environmental sustainability................................................................................... 22 3.2.2 Social sustainability ................................................................................................ 23 3.2.3 Economic sustainability .......................................................................................... 23 4 Device classes ....................................................................................................................... 24 4.1 Energy neutral device class .......................................................................................... 25 4.1.1 Key characteristics .................................................................................................. 26 4.1.2 Other characteristics ................................................................................................ 27 4.1.3 Technology enablers ............................................................................................... 28 4.1.4 System/infra enablers .............................................................................................. 29 4.1.5 Sustainability considerations................................................................................... 32 4.1.6 Key performance parameters .................................................................................. 32 4.1.7 Link to use-cases ..................................................................................................... 32 4.1.8 Example devices ..................................................................................................... 32 4.2 Reliable high data rate with bounded latency devices ................................................. 33 4.2.1 Key characteristics .................................................................................................. 33 4.2.2 Other characteristics ................................................................................................ 35 4.2.3 Technology enablers ............................................................................................... 36 4.2.4 System/infra enablers .............................................................................................. 36 4.2.5 Sustainability considerations................................................................................... 36 4.2.6 Key performance parameters .................................................................................. 37 4.2.7 Link to use-cases ..................................................................................................... 38 4.2.8 Example devices ..................................................................................................... 38 4.3 Highly reliable low latency device class ...................................................................... 39 4.3.1 Key characteristics .................................................................................................. 39 4.3.2 Other characteristics ................................................................................................ 40 4.3.3 Technology enablers ............................................................................................... 40 4.3.4 System/infra enablers .............................................................................................. 41 4.3.5 Sustainability considerations................................................................................... 41 4.3.6 Key performance parameters .................................................................................. 42 4.3.7 Link to use-cases ..................................................................................................... 42 4.3.8 Example devices ..................................................................................................... 43 Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 6 / 61 4.4 Enhancements of mMTC device class. ........................................................................ 43 4.4.1 Key characteristics .................................................................................................. 45 4.4.2 Other characteristics ................................................................................................ 46 4.4.3 Technology enablers ............................................................................................... 47 4.4.4 System/infra enablers .............................................................................................. 48 4.4.5 Sustainability considerations................................................................................... 48 4.4.6 Key performance parameters .................................................................................. 49 4.4.7 Link to use-cases ..................................................................................................... 49 4.4.8 Example devices ..................................................................................................... 49 5 Conclusion and Outlook ..................................................................................................... 50 6 Appendix A: Sub-THz transceiver design ........................................................................ 53 6.1 Design methodology .................................................................................................... 53 6.2 Scenario parameters ..................................................................................................... 54 6.2.1 Application requirements ........................................................................................ 54 6.2.2 Band of operation and regulations .......................................................................... 54 6.2.3 Propagation channel ................................................................................................ 54 6.2.4 Transceiver architectures ........................................................................................ 55 7 Appendix B: Companion devices ....................................................................................... 57 8 Appendix C: Example of novel device enabling a 6G capability .................................... 59 Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 7 / 61 List of Tables Table 3-1: Characterization criteria ................................................................................................................. 19 Table 4-1: Device classes overview ................................................................................................................ 24 Table 4-2: Typical energy sources for EN IoT devices, and corresponding characteristics and potential use cases (cf. [LRR+23] and references therein). .................................................................................................. 26 Table 4-3: KPIs for reliable high data rate with bounded latency devices [38.838] [HEX22-D13]. .............. 37 Table 5-1: An overview: 5G device classes against 6G device classes ........................................................... 51 List of Figures Figure 1-1: Hexa-X-II generic future device concept...................................................................................... 11 Figure 1-2: Methodology to identify 6G device classes in Hexa-X-II ............................................................ 12 Figure 2-1: Initial set of use-cases as described in Hexa-X-II [HEX223-D11] ............................................... 13 Figure 2-2: Use-cases and some example devices realizing them. .................................................................. 17 Figure 4-1: Placement of EN devices in the IoT landscape ............................................................................. 26 Figure 4-2: Role of an Infrastructure enabler device as system enabler for EN devices. ................................ 30 Figure 4-3: Using cellular infrastructure to read EN backscatter modulated messages. ................................. 31 Figure 4-4: Different XR device types. ........................................................................................................... 38 Figure 4-5: Evolution of cellular IoT and respective devices from 4G to 6G ................................................. 45 Figure 4-6: Example mMTC type devices ...................................................................................................... 49 Figure 5-1: Summary of 6G device classes ..................................................................................................... 51 Figure 6-1: Tap-delay response where successive taps represent different propagation paths between the transmitter and the receiver, delayed based on the path length divided by the speed of light. ........................ 55 Figure 7-1: Companion devices and network controlled flexible topologies. ................................................. 57 Figure 8-1: Network senses/detects a passive object using sensing signals and communication signals originating from a EN device which is linked to the passive object................................................................ 59 Acronyms and abbreviations Term Description ADC Analog-to-Digital Converter AGV Autonomous Guided Vehicle AI Artificial Intelligence AR Augmented Reality CaaS Compute-as-a-Service Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 8 / 61 CE Coverage Enhancement CPU Central Processing Unit CSI Channel State Information DAC Digital to Analog Converter DL Downlink DoF Degrees of Freedom DRX Discontinuous Reception D2D Device to Device eGPRS enhanced General Packet Radio Services EH Energy Harvesting EIRP Equivalent Isotropic Radiated Power eMBB Enhanced Mobile BroadBand EeMBB Enhancements of Enhanced Mobile BroadBand EmMTC Enhancements of Massive Machine-Type Communication EURLLC Enhancements of Ultra-Reliable and Low Latency Communications EMC Electromagnetic Compatibility EMG ElectroMyoGraphy EN Energy Neutral E2E End-to-End FDD Frequency Division Duplexing FOV Field of View FR Frequency Range GPU Graphic Processing Unit HD-FDD Half-Duplex Frequency Division Duplexing HMD Head Mounted Display HRLL High Reliability and Low Latency IIoT Industrial Internet of Things IoT Internet of Things JCAS Joint Communication and Sensing KPI Key Performance Indicator LEO Low-Earth Orbit Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 9 / 61 LOS Line Of Sight LPWAN Low Power Wide Area Network LTE Long Term Evolution MCL Maximum Coupling Loss MIMO Multiple-Input Multiple-Output ML Machine Learning mMTC Massive Machine-Type Communication mmW Millimeter Wave MQTT Message Queuing Telemetry Transport MTC Machine type Communication NB-IoT NarrowBand IoT NLOS Non Line Of Sight NLP Natural Language Processing NoC Network on Chip NR New Radio NTN Non-Terrestrial Network OFDM Orthogonal Frequency-Division Multiplexing OS Operating System PB Power Beacon PDB Packet Delay Budget QAM Quadrature Amplitude Modulation QoE Quality of Experience QoS Quality of Service RAN Radio Access Network RAT Radio Access Technology RedCap Reduced Capability RF Radio Frequency RFID Radio-Frequency Identification RHDRBL Reliable High Data Rate with Bounded Latency RIS Reconfigurable Intelligent Surface RRC Radio Resource Control Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 16 / 61 2.1.6 Interactive and cooperating mobile robots This use case illustrates how 6G will enable robotic systems to collaborate with each other and/or with humans in a reliable, autonomous way to achieve their common goal [HEX21-D12]. Such collaborative robot swarms can be deployed both outdoor (e.g., construction sites, agricultural plantations, etc.) as well as indoor (e.g., consumer homes, warehouses, production/manufacturing environments, etc.) [HEX22-D13]. As described in [HEX22-D13], on-premises 6G infrastructure must provide low latency (0.5 – 25 ms) communication to static and mobile (<10 m/s) automated guided vehicles (AGVs), UAVs and human operators to allow for reliable and safe interaction. To overcome environmental constraints such as Line-of-Sight (LoS) blockages, multiple frequency bands must be supported, i.e., sub-6 GHz bands for campus-wide coverage and higher frequency bands for local D2D communication. As safety is a crucial requirement, especially when interacting with humans, robots must be aware of their own position and the position of their peers. Therefore, accurate positioning (1 – 5 cm) must be supported at a 10 to 10000 Hz refresh rate. When involving UAVs, e.g., for warehouse inventory management, these challenges become increasingly difficult as 3D positioning is required. 2.1.7 Small coverage, low power micro-network in networks This use case, with the full title “Small coverage, low power micro-network in networks for production & manufacturing” detailed in [HEX21-D12] and [HEX22-D13], relates to the envisioned 6G use case family of trusted embedded networks where high-level trustworthy communication capabilities of very sensitive information are required for sub-networks, or networks of networks, tightly integrated in wide-area networks. It considers the application where a machine manufacturer wants to mutually connect a large population of sensors in the machine using non-Industrial, Scientific, or Medical spectrum for reliability reasons. As described in [HEX21-D12], this can be seen as a shared spectrum access concept with infrastructure-less networks of very limited coverage as an underlay network under full control of the incumbent, involving large machine population interconnected via very low-power devices (resilient IoT sensor devices), potentially with one of the sensors getting the authorization out of the public or non-public network of which the spectrum is used (Trustworthy/Intelligent aggregator sensor node or smart IoT gateway device). Regarding deployment characteristics, the typical environment will include rural areas as well as indoor case (e.g., production site) [HEX22-D13]. Very low-power devices will be needed to mutually connect sensors mounted in manufacturer’s machine or vehicle, with possibly high density of tens to some thousands of sensors (few sensors per cubic meter for manufacturing case). These sensors can be static (e.g., environment sensors) or mobile (when e.g., mounted to a possibly low mobility vehicle). Standard licenced sub-6 GHz frequency bands (shared with the infrastructure) can be assumed for low power consumption, but higher frequency bands might be used as well. The main expected technical challenges originating from environmental or other constraints include electromagnetic compatibility (EMC) requirements (e.g., for sensor-to-sensor connectivity with proximity), small form factor and low-cost, and safety of operation requirements. Capabilities of data security (for protecting the processed proprietary information and guaranteeing the availability of the connectivity, independent of the infrastructure network), interference mitigation (due to dense deployment and flexibly shared spectrum), and localization in several cases, will be of paramount importance. 2.2 Services 2.2.1 Compute as a Service Compute-as-a-Service (CaaS), which is a service described in [HEX21-D12] and [HEX22-D13], is also analyzed as some of the low capability devices need to rely on this service to augment their compute capabilities. CaaS is more of a use-case enabling service and could enable devices, especially resourceconstrained ones, to delegate resource intensive processing tasks. For example, CaaS may be used for storing and processing of sensory data collected from special equipment such as glasses, gloves vests etc in an industrial setting. Another example could be in multi-player gaming where complex games may be processed on computational resources in the network. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 17 / 61 Thus, typical deployment of CaaS will be for devices with limited compute capability that could be operating in both indoor and outdoor environment. Such devices could be sensors that are static or wearables that would have mobility. These devices could be operating on FR1(sub-6 GHz), FR2 (mmWave) and potentially future ones in centimetric range as well as sub-THz as dependent on the application/use-case. To deploy CaaS, some of the notable technical challenges that should be addressed are discovery of suitable CaaS providing entity and therein ascertaining available compute resources, trustworthiness, energy footprint etc., decision on when and where to offload (based on current/predicted resource availability and performance of the service providing entity) and simple, generic interfaces that are abstracted from underlying heterogenous hardware (e.g., Central Processing Unit (CPU), hardware accelerators etc.) and software (e.g., Operating System (OS)) as much as possible to request/execute the service. Though depending on exact application or usage scenario it can be anticipated that devices needed to support high availability, high reliability, low latency in addition to overall improved energy efficiency while ensuring high security and user data privacy. For example, AR glasses offloading resource intensive tasks to another capable trusted node would require reliable low latency communication to the service providing node, which can ensure user data privacy. 2.2.2 Situation aware device reconfiguration Situation aware device configuration, as described in [HEX22-D13], represents a transformative service poised to enhance the capabilities of interconnected devices. This service empowers devices with the agility to dynamically alter their device types and configurations to align seamlessly with evolving service requirements. Device endowed with this service gain the remarkable ability to transition between different device types, harnessing diverse processing capacities and functionalities based on the precise demands of the active service. This dynamic metamorphosis ensures that the device is optimally configured to deliver peak performance for the task at hand. This service can not only respond to existing requirements but also predict future demands. These predictive insights enable devices to anticipate the need for a change in device type and timing, ensuring a proactive response that guarantees uninterrupted service excellence. The challenge for 6G systems lies in the extremity of performance requirements for new applications, for the addressment of which current 5G solutions may need to be enhanced. Flexible device type change may evolve today’s network slicing concept in 5G to the one of device slicing. In one instance of the usage scenario of situation aware device configuration service, a familiar consumer device like a smartphone seamlessly shifts roles upon entering Vehicle-to-everything (V2X) communication zones, transitioning from its usual voice/data function to become a conduit for safety-related communication, ensuring warnings for vulnerable road users and facilitating vehicle-to-network services. In another scenario, an industrial robot adeptly toggles between critical tasks like high-precision welding demanding minimal latency and exceptional localization accuracy, and less critical tasks involving longer-range movement with more moderate precision and latency requirements. These examples showcase the dynamic adaptability and performance optimization enabled by this innovative service, poised to redefine device interactions. 2.3 Summary of motivating use-cases and services analysis Figure 2-2: Use-cases and some example devices realizing them. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 18 / 61 One key takeaway after analysis of various use-case and services is that one use case may need multiple types of devices to realize it. The other angle is also relevant – a device could serve multiple use cases. For example, energy efficient sensors that rely on energy harvesting as an energy source could play a contributing role in Earth monitoring as well as in Immersive smart cities. Another notable takeaway is that one device type could have varying implementations depending on the use-case it serves. For example, an IoT device could have integrated circuit design signoffs done for different temperature range for consumer use-case deployments than one that should be deployed in an industrial scenario. Another aspect could be the placement of antenna: in some industrial robots it might be placed close to the device’s radio front end components while in some other devices it might be placed remotely to enhance reception. Furthermore, if a device must operate in a low coverage or have high mobility then it may be designed with only supporting low frequency spectrum. Thus, a device design could be adapted to the use-case it is serving and its deployment scenarios. Furthermore, a single device can have multiple capabilities. As seen from Section 2.2.2, a device may operate in one operating mode where it is catering to one service and then in another operating mode where it caters to another service. Such devices could be dynamically re-configured as well, but in essence it belongs to one device class/type as all features need to be implemented. In one operating mode some features (e.g., sensing) are more used or even enhanced than other features (e.g. communication). Along the similar lines, a sensing and communication device utilizing the same radio spectrum could be seen as a multi-mode device and not as a new class of device. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 19 / 61 3 Characterization criteria and sustainability guidelines This chapter outlines the characterization criteria that sets the basis to understand if the different devices in the 6G timeframe fall into a one device class or another. Furthermore, the sustainability guidelines from [HEX223D11] are summarized in chapter 3.2 to enable identifying the sustainability considerations for each identified device class. 3.1 Characterization criteria There are numerous characteristics concerning a device, such as its energy consumption, operating spectrum, authentication etc. Table 3-1 below contains the characteristics that form the characterization criteria, along with some description that also provides some granularity for each of these characteristics. The related characteristics are grouped, and the group name is indicated in the column titled “Group”. The granularity is set at a coarse level to serve as a guide and depending on the exact implementation etc., it is possible that there are some sub-levels of devices which depict same or very similar high-level characteristics. Table 3-1: Characterization criteria Group Characteristic Description Comments Energy Energy source Energy source for the device: {generated ondevice (for e.g., energy harvesting based), generated remotely (e.g., power grid charging a local on-device battery)} Energy storage Identify storage type {Battery, super capacitor, None}. Indicate storage material with environmentally material or not Identify if it enables zero eWaste. Energy consumption during (1) operation (2) idle/sleep Distinguish between active operation and sleep power. Difficult to have exact numbers as they are use-case and implementation dependent indicate a range (in mW) Lifetime Device lifetime Indicate if the device has a lifetime that is within a cellular network generation (Nth G), or its deployment can span multiple network generations (for e.g., Nth G, (N+1) th G This is dependent on the usage scenario (e.g., discardable smart tags, infra sensor). Device design/manufacturing aspects such as battery lifetime (if irreplaceable), material lifetime, etc. could also influence. Mobility Device mobility Fixed device or Mobile {low (human powered), medium/ (car, train <200 kmph), high (airplane >200 kmph} Impact on radio propagation, fault fixing, life cycle management. Indicate typical speeds based on the deployment scenario. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 20 / 61 Communication Authentication Distinguish between light weight/device energy aware authentication or typical authentication. All devices shall be uniquely identifiable on the network. Synchronization Distinguish if the device supports Timing as a Service Time aware system component Distinguish if the device has time aware system components (e.g., Device Side-TSN Translator in a TSN) Identify if the network is a part of a bigger time-aware system. E.g., Time Sensitive Network based IoT system. Spectrum Distinguish between new (cmW, sub-THz) and existing spectrum (FR1 and FR2) & licensed and unlicensed Mainly identify impact on the front-end RF and antenna components. Traffic flow Distinguish between {sporadic/aperiodic, energy source dependent traffic flows, Quasi periodic, periodic, ...} Identify the impact on the latency and power consumption Data rate Distinguish between the typical data rates that are supported. Latency Distinguish between the bounds on the latency- {none, time critical/bound, URLLC, …} Reliability Distinguish between levels of reliability at a coarse level: {none, exists, stringent} Percentage of the amount of sent packets delivered within QoS constraints. Availability Availability requirements to be supported. NTN support Yes/No Indicates if the device supports Non-Terrestrial Network (NTN) in addition to TN. And if supported, indicate the services supported: is it for only emergency call, or also data call etc. AI & computation Computation capability Distinguish between {Low, medium, high, AI-enabled, offloading dependent,} Identify computational capability and dependence on other entities. Low computation capabilities indicate simple circuits, micro-controller-based Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 21 / 61 designs etc., while Medium indicates CPUs, low end GPUs etc. High end computation capabilities would have additional dedicated HW accelerators, high end GPUs, CPUs etc. AI-enabled is an additional qualifier for identifying devices running ML kernels, while offloading dependent indicates devices that need another more capable remote compute to perform its functionalities. Localization & sensing Location accuracy Distinguish if the device type has requirements on location accuracy and if yes, in horizontal & vertical position accuracy (in m) Orientation accuracy Distinguish if the device type has requirements on orientation accuracy and if yes, for roll, pitch, yaw (in degrees) Localization/sensing latency Distinguish if the device type has latency requirements w.r.t. localization/sensing service latency and if yes, indicate granularity of requirements {low, medium, high} latency. Security Security capability Security (e.g., encryption, data integrity) features supported on device. Distinguish between {none/minimal, medium secure, highly secure, specialized} This parameter is about the security features supported on the device and indicates the potential vulnerabilities/risks. None/minimal level is for devices such as low-cost IoT sensors or older devices that have minimal/no built-in security features such encryption, secure boot, no updates etc. Medium secure level is for devices that have some amount of security features such as single level depth of defense, Trusted Execution Environment (TEE) etc. but Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 22 / 61 have less features than highly secure devices. Highly secure level for devices that assume zero trust and support advanced features such as hardware root of trust, multiple levels of depth of defense, dedicated secure vaults etc. [HLN20]. Furthermore, the highly secure devices could undergo rigorous testing etc. as these devices could be part of the critical infrastructures and support specialized features such as permanently de-activate a device that is de-commissioned from the 6G system, dedicated processing solutions could be cryptographic acceleration sub-systems, secure Network on Chip (NoC)/interconnects. 3.2 Sustainability guidelines As part of the effort to creating a sustainable future, the device classes also indicate some of the important sustainability considerations to be taken into account during each device class design. The sustainability considerations are based on the sustainability guidelines provided in [HEX223-D11]. These guidelines are organized under environmental sustainability, social sustainability and economic sustainability and are detailed in the following sub-sections. 3.2.1 Environmental sustainability To create a more sustainable and resilient communication system, 6G technology design need to incorporate environmental sustainability considerations in the planning, architecture and operation of the network, including: • Holistic footprint approach – 6G network will have different footprints and thereby different environmental impacts. A holistic and comprehensive approach to different footprints, taking into account all the trade-off between different sustainability factors is necessary. • Alternative materials – Hazardous and chemical substances should be avoided/restricted in the production of 6G networks. This can be done by complying with the existing regulations on restriction of hazardous substances and chemicals and using alternative materials instead. • Modular and durable equipment – Modular design of 6G allows for extended lifespan and easier customization and scalability of the system while reducing waste generation and downtime. • Energy Efficiency – A significant increase in future traffic is expected which subsequently can lead to higher power consumption. Energy efficiency is therefore of importance in the design of 6G technology, in both production and operational phases. Strategies to direct towards energy efficient 6G include the use of low-power components, optimization of network architecture and protocols, implementation of energy-efficient algorithms and use of more renewable energy resources. • Cloud Computing and Automation – Automated and intelligent system which help to reduce the amount of data or power consumption in the system needs to be considered. For instance, by Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 23 / 61 moving some of the processing power from the cloud to a location to the end user or the device, through edge computing, total amount of data and total energy consumption of a 6G system can be reduced. Another example is to manage the on/off states of networks and devices to match the demand of the system. • Circularity practices – Achieving circularity require a fundamental shift in how equipment and devices are designed, produced, used, and disposed of and needs actions from all stakeholders. 6G system needs to adopt circularity practices for instance via the use of renewable energy resources, design of equipment that can easily be recycled or disassembled at their end-of-life and design of equipment that that their components can be reused, refurbished, recycled, and repaired. 3.2.2 Social sustainability Social sustainability is comprised of two main aspects: Trustworthiness and digital inclusion. The following social sustainability aspect needs to be considered when designing 6G technology. • Cyber-secure and respect end-users’ privacy – 6G solutions, devices and networks need to be cybersecure and respect end-users’ privacy. • Transparent AI-based approaches – AI-based approaches need to be clear, transparent. It should also keep the human in the loop. These aspects enable accountability, and thus trust, can be maintained. • Flexible capacity and coverage – 6G networks need to adjust capacity and coverage depending on geographical areas and the end-users’ needs. It is also important that the 6G networks is flexible to adjust to the criticality of the offered solutions, e.g., e-health aspects vs. entertainment. • Non-discriminatory – 6G solutions, including devices, need to for e.g., consider IT literacy and culture of all types of end-users. • Ensure coverage and capacity at certain level – This helps to maintain people’s trust in digital capabilities and services. 3.2.3 Economic sustainability 6G technology design needs to also incorporate economic sustainability considerations. The following drivers and goals are identified from economic sustainability perspective for 6G. • Value-based 6G design – 6G should be designed such that it brings economic value beyond users/buyers. It should provide opportunity for different stakeholders to harness converging technologies to create an inclusive, human-centred future. • Sustainable 6G business model innovation – 6G should be developed to solve major sustainability challenges while being sustainable. Sustainability will be the source of value and sustainability can become the core of future business models. • Open value configurations via 6G – 6G needs to be developed encouraging innovation via open value configurations. It should share economy models which allows stakeholders to innovate and create new collaborative business models for the whole 6G ecosystem while considering security concerns. • Sustainable competitive advantage via correlated/holistic sustainability perspective in 6G – 6G development needs to identify the complex interdependencies and trade-offs between the three sustainability pillars, i.e., environmental, social, and economic sustainability. This is to develop a new correlated sustainability perspective to reach long-term competitive advantage in the new 6G ecosystem. • Monetizing with 6G in challenging business environment – 6G needs to bring return on investment in different environments including supporting the varying density of users, data, and energy usage. As this monetization should be achieved while operating under increasing environmental sustainability requirements, different business models are called for. • Preparing for mitigating risks with 6G – By 2030, the time of launching 6G, environmental risks such as climate action failure as well as social risks such as geoeconomic confrontations, digital inequality and cybersecurity failure will have potentially big impact on the economy of nations, companies, and individuals. It is therefore important that 6G development and deployment prepare for protecting stakeholders from the increasing risks in the changing operational environment related to environmental, social, and business aspects among others. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 24 / 61 4 Device classes From the use-case analysis in Chapter 2, it is evident that a single use-case may be served by multiple devices and possibly belonging to different device classes. In this document, most of the analysis is based on the [HEX223-D11] use cases. The upcoming Hexa-X-II deliverables, such as D1.2, aims to propose an enhanced methodology for analyzing the use cases and identifying the limitations of current technology etc. This new analysis could have an influence on the identified device classes in this document, resulting potentially in new device classes such as the ones related to the enhancements of URLLC and eMBB types of devices in 5G. This is noted in Chapter 5 and is planned along with joint sensing and communication aspects to be elaborated further in a later deliverable from the Work Package 5 and based on the upcoming Hexa-X-II deliverables. We have defined an exhaustive set of the characterization criteria in Chapter 3, which is organized into groups. Among the groups of characterization criteria, we attached more weight to ones from energy and communication/radio performance to determine distinct device classes, which is summarized in Table 4-1. This gives a good overview of the parameters that impact the energy and (radio) performance of the end-toend system. In Table 4-1, relative indications (very low – low – medium – high - ultra) are used to depict the distinction between the device classes. For latency, reliability and availability, the transition from medium to ultra is one order of magnitude tighter per each higher step. For e.g., ultra has 99.999% reliability while high would translate to 99.99% and medium would translate to 99.9% reliability. As noted in [HEX23-D14], the literature study indicates variations between the numbers for the mentioned criteria among similar use-cases and exact values depends on the use-cases, how the solution is architected etc. Table 4-1: Device classes overview Device class name Energy Neutral (EN) device class Reliable High Data Rate with Bounded latency (RHDRBL) device class High Reliability & Low Latency (HRLL) device class Enhancements of mMTC (EmMTC) device class. Criteria Energy During operation, Energy neutral (very low energy consumption) Low energy (lowering energy consumption could take precedence over reliability) As low energy consumption as possible (without compromising on reliability) Low energy consumption Data rate Very low High Medium Low Latency No mandatory requirements Bounded latency Low latency No mandatory requirements Reliability No mandatory requirements Medium High Low Availability No mandatory requirements Medium High Low In Table 4-1, Energy Neutral (EN) device class represents very low data rate devices that function based solely on harvested energy. Zero energy devices, a device type that is identified in the use-case section, are mapped into energy neutral device class. Reliable High Data Rate with Bounded Latency (RHDRBL) device class refers mainly to devices that enable immersive experiences such as XR devices. These class of devices needs high data rates and bounded latency (= consistent low latency) more in the order of tens of milliseconds yet at some amount of reliability (for e.g., 99.9%). These reliability requirements are not as high as the High Reliability and Low Latency (HRLL) class of devices, which is mainly about devices in the Industrial IoT domain. The devices in the HRLL device class would typically need reliability around 99.99% and latency of about 1-10ms range. Therefore, they have more relaxed requirements than the Ultra Reliable and Low Latency (URLLC) devices in 5G, which are typically defined to have 99.999% reliability at 1 ms latency for 32 bytes packets [LSK20]. Finally, the enhancement of mMTC device class aims to provide enhancements to Low Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 25 / 61 Power Wide Area Network (LPWAN) devices, focusing mainly on power consumption, slightly higher data rates than in the 5G time frame, and maybe on the reliability requirements. It is good to note that some indicative numbers are available within the respective device class sections and depend on the specific deployments. The rest of this chapter is organized as follows. Each of the identified device classes is elaborated for their key characteristics that distinguishes it from any other device class (e.g., operates only on harvested energy), other characteristics that are similar to another device class (e.g., low data rate that both EN and IoT sensors have in common) and their technology enablers (technologies that help to realize devices within that device class). Furthermore, system / infrastructure enablers (if any) that will support the devices to be integrated in a 6G system and sustainability considerations that could be relevant to a class of devices is also described. Where possible to have consistent numbers backed with references, the key performance parameters are detailed with these numbers. Finally, a link to the use cases for each device class and examples of devices that belong to a specific class of devices are provided. 4.1 Energy neutral device class Energy Neutral (EN) devices are energy harvesting IoT devices that embody an ultra-low energy consumption and demand throughout the entire life cycle (from manufacturing to disposal), facilitate zero waste generation, and embraces material circularity principles. The key energy-related principles of EN IoT devices are: • Energy-efficient design and operation: Operation must be designed to minimize energy consumption, e.g., by optimizing hardware components, reducing power requirements, and utilizing lowpower/sleep modes. • Energy-conscious manufacturing: The manufacturing must leverage green practices and technologies, e.g., using energy-efficient/green equipment, optimizing production processes, and low-cost, circularity, and biodegradable materials. • Energy-aware deployment: The deployment must maximize energy efficiency considering short and long-term impacts. For instance, the devices may be deployed such that their exposure to energy sources for harvesting and their sensing/transmissions capabilities are properly balanced considering time-varying environment/network/device conditions (e.g., seasons, coexisting networks/devices, hardware aging from the environment, network, and device perspective, respectively). • Energy-frugal disposal: Recycling, repurposing, and biodegradability practices must be promoted to minimize energy consumption associated with disposal and reduce waste generation. Zero-energy devices, a device type identified in Chapter 2 maps to this device class. Some deployment scenarios include but are not limited to asset tracking, transportation & logistics, warehouse, industrial (e.g., factory automation, harbors, docks), stores (e.g., automatic inventory), or smart home. Figure 4-1 depicts the placement of EN devices in the IoT device landscape. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 32 / 61 4.1.5 Sustainability considerations The sustainability considerations for the EN device class are grouped under the following three sustainability pillars: Environmental sustainability • Ultra-low energy consumption and demand throughout the entire life cycle, from manufacturing to disposal. • Using renewable energy sources and biodegradable or recyclable materials whenever possible. Social sustainability • Promote digital inclusion and well-being by providing enhanced wearable and inor on-body sensor and corresponding services. • Social equity and well-being by providing for instance safety and security measures under environmental hazard and security condition. • Avoid data leaks, and thus preserve data privacy. Economic sustainability • Optimizing resource usage and efficiency resulting in improving maintenance procedure. 4.1.6 Key performance parameters The following are the key performance parameters for the EN device class. Ultra-low energy consumption The primary sources of energy consumption typically include wireless communication (transmitting and receiving data), computation (processing data and executing algorithms), and sensor operation (collecting and processing environmental data). EN devices may have limited capabilities in terms of communication, computation, and/or sensing to limit energy consumption, costs, and form factors. The existing capabilities must be optimized, along with managing sleep modes and duty cycles, to achieve overall energy efficiency. Highly efficient energy harvesting Energy-harvesting capabilities may be location-dependent, as the availability of the energy sources may vary accordingly. Therefore, in very simple systems, deployment optimization is crucial. In more complex systems, efficient energy management techniques may suffice. In any case, the energy harvesting circuits must be optimized for maximum efficiency. Efficiency degradation over time, e.g., due to the degradation of materials, wear and tear, or contamination, must be considered. Low energy consumption per successful transmission EN devices must be engineered to operate efficiently in environments with a high density of interconnected nodes, ensuring minimal interference and spectral occupancy. For doing so, they must employ lightweight and energy-aware communication protocols, spectrum utilization techniques, and channel access mechanisms. EN IoT devices may contribute to the establishment of scalable networks, enabling widespread deployment and effective collaboration among a multitude of devices while maintaining stringent power constraints. Therefore, in addition to data transmission at low energy consumption, leveraging energy efficient/aware MAC or similar technologies, such transmissions should have a high probability of success in a dense environment. 4.1.7 Link to use-cases EN devices assist many use-cases, such as, E-health for all, autonomous supply chains, earth monitoring. The devices can work in conjunction with other IoT, NTN, eMBB and XR devices to support the uses cases. The requirements of use case can influence the EN device integration - passive, semi-passive or active EN device. 4.1.8 Example devices The devices can appear with varying form factor and hardware/electronics (backscattering or harvesting or combination based). Examples are tags, stickers, inor on-body sensors. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 33 / 61 4.2 Reliable high data rate with bounded latency devices The reliable high data rate with bounded latency (RHDRBL) device class provides immersive experience, through mixed, augmented, and virtual reality. With high guaranteed data rate, reliability and bounded latency demands, it ensures quasi-realistic virtual experience for merging cyber and physical worlds. This class supports diverse communication needs and encompasses different functions such as visual devices/headsets/glasses as well as wearables and haptics, ensuring seamless integration with various devices. The following are some definitions of the main devices in this device class [26.928], [38.838]: • Extended reality (XR): represents real-and-virtual environments combinations as well as interactions between humans and machines. It is a generic term that encompasses for AR, MR and VR which will be explained later. It also considers all the fields interpolated between these terminologies. The degree of virtuality cover a wide spectrum from partially sensory inputs to fully immersive VR. The main goal of XR is to extend the experiences of humans regarding the senses of existence (which can be achieved by VR) and the acquisition of cognition (which can be achieved by AR). • Virtual Reality (VR): given some delivered audio and video data, a virtual reality rendered version is delivered to the user. This rendered version is designed to quasi replicate the visual and audio sensory data received from the real world to a certain limit defined by the application. VR applications often require the user to use a head mounted display (HMD), to replace the user field of view with the rendered visual components instead of real world. Also, headphones are needed to enrich the VR experience with accompanying audio. Moreover, head and motion tracking sensors provide better reactivity for visual and audio components to be accurately updated and synchronized with the VR user movements. • Augmented Reality (AR): represents the overlay of additional artificially generated items or contents on top of the real environment. This additional content is mainly a visual and/or audio content that can be directly relative to an observation from the surrounding or indirectly through additional perception of the environment through sensing. In the case of indirect reaction to the environment, additional sensing, processing, and rendering may be needed. • Mixed Reality (MR): is an evolutionary version of AR in which some elements can be added virtually into the real environment that leads to an illusionary view that these elements exist physically in the real environment. 4.2.1 Key characteristics Device class where applications need high (guaranteed) data rate with bounded latency; high availability and device orientation & location accuracy influences QoE. Quality-of-Experience for XR devices mainly depends on the following factors which are immersiveness and presence and interaction delays and age of content, which can be explained in more details as follows [26.928], [38.838]: Immersiveness and Presence: immersiveness and presence encompasses the sensation of being physically and spatially situated within a particular environment. This concept can be categorized into two distinct types: • Cognitive Presence: pertains to the presence of one’s mental faculties. It can be attained through activities such as engrossing film-watching or captivating book-reading. Cognitive presence holds significance for creating an immersive experience in various contexts. • Perceptive Presence: relates to the involvement of one’s sensory perception. Achieving perceptive presence involves the manipulation of one’s senses, including sight, sound, touch, and smell. To establish perceptive presence, XR devices must effectively deceive the user’s sensory perception, particularly the visual and auditory systems. XR devices accomplish this by employing positional tracking based on the user’s movements. The primary objective of this system is to maintain the user’s sense of presence without disrupting it. Presence is attained when the instinctual elements of our primitive brain regions are triggered. There are four key elements that contribute to the sensation of being present: • The illusion of being in a stable spatial place Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 34 / 61 • The illusion of self-embodiment • The illusion of physical interaction • The illusion of social communication The most crucial aspect from a technical standpoint is the first element. This facet of presence can be subdivided into three general categories, ranked in order of their significance in shaping presence: • Visual presence • Auditory presence • Sensory or haptic presence Visual presence requires the following: • Tracking: 6DoF tracking (where 6DoF or 6 degrees of freedom, represent the 3D position and the rotation of the device), 360 degrees tracking (where 360 degrees tracking defines the ability to track the user’s head regardless of the direction the user is pointing at), sub-centimeter accuracy, quarterdegree-accurate rotation tracking, no jitter, comfortable tracking volume, above 1000 Hz tracking frequency (where tracking frequency defines how often the device needs to relate its current location to a prior one for accurate tracking). • Latency: < 20 ms motion-to-photon latency, where motion to photon latency defines the delay from the time you move your head to when you see the display change. To meet that 20 ms, it does not necessarily mean the 6G round-trip needs to be extremely low. Intelligence on the device can help achieving these using techniques such as asynchronous time warping and space warping. • Persistence: Low persistence - Turn pixels on and off every 2 - 3 ms to avoid smearing / motion blur; 90 Hz and beyond display refresh rate to eliminate visible flicker. • Resolution: • Spatial Resolution: No visible pixel structure - you cannot see the pixels. • Temporal Resolution: a sustained frame rate of 90 Hz and beyond. • Optics: Wide Field of view (FOV) is the extent of observable world at any given moment and typically 100 - 110 degrees FOV is needed; Comfortable eyebox (eyebox defines the volume where the eye receives an acceptable view of the image); High quality calibration and correction - correction for distortion and chromatic aberration that exactly matches the lens characteristics. The sense of presence holds significance not only in VR but also in immersive AR. To establish a sense of presence in AR, it is crucial to seamlessly blend virtual elements with the physical environment. Just as in VR, the virtual elements must align with the user’s expectations. In the context of truly immersive AR, especially MR, the goal is for the user to be unable to distinguish between virtual objects and real ones. Noteworthy for both VR and AR, with particular emphasis on AR, is the importance of considering not just the user’s experience but also the awareness of the surrounding environment. This encompasses: • Safe zone discovery • Dynamic obstacle warning • Geometric and semantic environment parsing • Environmental lighting • World mapping In the realm of AR, to improve their perception of the actual surroundings, individuals might utilize HMD to observe three-dimensional computer-generated objects overlaid onto their real-world perspective. This seethrough functionality can be achieved through either an optical see-through or a video see-through HMD. Interaction Delays and Age of Content: Beyond the aspects of presence and immersion, the age of content and the delay in user interaction are paramount considerations for both immersive and non-immersive interactive experiences, such as those found in online gaming. The user interaction delay, age of content, and round-trip interaction delay, are defined as follows: • User interaction delay: refers to the time elapsed between when a user initiates an action and the when the content creation engine acknowledges and incorporates that action. In the gaming context, Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 35 / 61 this delay represents the time between a player’s interaction with the game and when the game engine processes and responds to the player’s input. • The age of content: on the other hand, signifies the duration between when the content is initially generated and when it is presented to the user. In the context of gaming, this duration encompasses the time between the creation of a video frame by the game engine and when that frame is ultimately displayed to the player. • The round-trip interaction delay: it comprises the combination of both the age of content and the user interaction delays. 4.2.2 Other characteristics Device class where applications need high requirements on computation, traffic and benefits from AI enabled compute engine, given device limited form factor. XR traffic has strict QoS requirements: XR traffic has strict QoS requirements such as bounded-latency, high data rate and reliability requirements. XR traffic requires strict motion-to-photon latency - <20 ms of overall latency (motion to photon latency defines the delay from the time you move your head to when you see the display change). Also, XR traffic requires significantly high data rates that can reach few Gbit/s for DL flows while maintaining reliability constraints of 99.9%, which means that the loss rate of packets transmitted end-to-end (E2E) across a network should be less than 0.1%. XR traffic frame multiple quasi-periodic DL and UL flows: The XR traffic is represented by a sequence of frames arriving at a receiver node from a transmitter node, according to a given traffic frame rate/periodicity. It also has a random variation in time which is defined as jitter. The size of each data frame, in the sequence of frames, also variates according to a certain random distribution. The typical XR traffic frame size and jitter are modelled by truncated Gaussian distribution [38.838]. XR traffic has variable frame sizes that can be modelled by a truncated gaussian distribution to capture the variation in bytes between different frames. Also, beside the XR traffic cadence and periodicity requirements, frames can arrive with some variation in time which is modelled by truncated Gaussian distribution as well. Moreover, XR typical traffic packet generation rate does not match the 6G slot granularity. The XR packet arrival rate is determined by the frame generation rate, e.g., 60 fps. Thus, the average packet arrival periodicity is given by the inverse of the frame rate, e.g., 16.6667 ms = 1/60 fps, where fps is the number of frames per second, and it represents the frame arrival rate. This non integer value of packet arrival rate causes the mismatch with 5G RAT slot granularity. The current NR slot granularity is dependent on subcarrier spacing (SCS) as follows: for SCS of 15, 30, 60, 120, 240 kHz, the corresponding slot durations are 1, 0.5, 0.25, 0.125, 0.0625 ms respectively. Thus, mechanisms defined in 6G need to be compatible with such cadence (for example with enhanced CDRX or other approaches). Computation (Standalone/direct link to network, via an (intelligent) aggregator) XR devices can apply local compute for processing/rendering at the device itself. However, this can induce higher latency, more power consumption and lower QoE. 6G capabilities can ensure high data rate – low latency communication links to the edge compute servers. Hence XR devices can leverage computing power from the edge compute server for graphics rendering, through connecting to these servers directly over 6G or through an intelligent aggregator. Device Capabilities: This device class is characterized by small form factor, limited battery size, expectation of long duration wearability. Some XR devices will not be capable of deploying four receive antennas at low frequency bands due to form factor limitations. Therefore, some two receive devices with co-polarized antennas will need to cope with the high data rate-bounded latency-high reliability requirements over 6G. This can be achieved by fixing the UE BW to 100 MHz for these 2Rx limited form factor devices and considering their special device characteristics and KPIs that need to be achieved over 6G. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 36 / 61 4.2.3 Technology enablers AI as a service AI and ML have made a significant impact on nearly every industry, pushing the boundaries of what high technology can achieve. In the context of XR services and delivering XR wirelessly, there are two crucial types of intelligence required to ensure effective and efficient performance: service intelligence and operational intelligence. Service intelligence is essential for tasks directly related to the XR application itself. This encompasses ML processes that handle rendering actions associated with the VR scene and the coordination of multiple holograms. On the other hand, operational intelligence equips the network with intelligent mechanisms to optimize and sustain complex services like XR, ensuring their smooth operation and selfsufficiency. Perception It facilitates enhancing the device awareness by applying inferencing methods applied to a collection of sensor and wearable data. For example, the perception is responsible for estimating the motion and the environment such that the metaverse content can be rendered to reflect the user’s movement and the environment. Spectrum It utilizes FR1, FR2, and potentially operate also in centimetric range and sub-THz bands for different objectives (capacity/coverage, data rate, etc.): A good XR user experience requires high data rates, highreliability, and low/ultra-low latency simultaneously. FR1, although reliable and ideal for mobility, cannot provide high data rates and ultra-low latency rates. The mmWave and sub-THz bands can fulfil data rates and latency requirements but are limited in range and mobility. It is good to note that considering these trade-offs along with the deployment scenarios would be helpful in optimizing sub-THz transceivers, whose design aspects are covered in Appendix A. Also, operating at centimetric range can provide a good trade-off between coverage and capacity. Energy optimized services It is required for users who want to add additional attention to energy consumption. 4.2.4 System/infra enablers Edge computing XR system can leverage computing power from the edge compute server for graphics rendering. With edge computing, user can access services hosted close to the serving 6G network entity. This approach helps to improve both end-user experience and network efficiency. Lower latencies can improve end-user experience, while reduced backhaul transport requirements can improve network efficiency. 4.2.5 Sustainability considerations The sustainability considerations for reliable high data rate with bounded latency device class are grouped under the following three sustainability pillars: Environmental sustainability • XR and other devices enabling fully merged cyber physical worlds enhances the collaborative work environments which can in turn reduce the need to vehicular mobility/traveling and thus, can be seen as an efficient way to reduce energy-related CO2 emissions (“climate action”). Social sustainability • Using Mixed Reality (MR) devices for virtual medical consultations can improve the access to healthcare and removes the distance barriers between the patients and the doctors (“access to quality health-care services”), • Improve teaching quality and availability through providing mixed reality access between teachers and students (aided by MR devices), while removing the distance barriers (“access to education”). Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 37 / 61 Economic sustainability • XR devices, for example, facilitate socializing with family and friends and thus reducing the need for traveling and increase the possibility to experience other countries more often without traveling there (“sustainable tourism”). 4.2.6 Key performance parameters Several performance metrics need to be satisfied for XR devices [38.838] such as: • Latency: The latency requirement of XR traffic is modelled as packet delay budget (PDB). The PDB is a limited time budget for a packet to be transmitted over the air from a 6G network entity to a user. XR devices require <20 ms PDB. • Cell Capacity: is defined as the maximum number of users per cell with at least Y % of UEs being satisfied. A UE is satisfied if all the considered streams meet their own PDB requirements, i.e., more than a certain percentage of packets are successfully transmitted within a given air interface PDB. To maximize the coverage of XR devices over 6G, cell capacity should be maximized. • Data Rate: XR devices requires significantly high data rates over 6G that can reach few Gbit/s for DL flows. • Power Consumption: XR devices require low power consumption over 6G to ensure longer battery lifetime, low thermal dissipations, and long duration of wearability. The following Table 4-3 indicates the values for some of the key performance indicators. Table 4-3: KPIs for reliable high data rate with bounded latency devices [38.838] [HEX22-D13]. KPI Target Value Reasoning/Reference Availability [%] 99% It defines the probability that the system is operational when a demand to access the service is made. It is calculated as: Uptime / Total time (Uptime + Downtime). This device class needs acceptable availability of 99%. Reliability [%] 99.9% It represents the packets’ loss rate which are transmitted end-to-end across a network. Strict reliability requirement of 99.9% is needed, yet more relaxed compared to URLLC. Latency [ms] <20 ms Combined E2E roundtrip for both UL and DL should be kept below 20 ms. E2E delay is defined as the time needed for a piece of data to be transmitted from a source point to a destination one across the network. Date Rate Up to 1 Gbit/s DL 0.1 Gbit/s UL It defines the typical required data rate to achieve a sufficient QoE for an application that uses this device class. Normally, the application should be able to adapt the data rate based on the radio condition and the loading of the network. Energy Consumption Dependent on resolution, frame rates and display technology. More details are explained in the rest of the subsection. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 38 / 61 Energy Consumption: In the process of designing media processing capabilities, XR functionality and integrating network connectivity, it is crucial to have a comprehensive understanding of the energy consumption associated with various components that might be incorporated into XR devices. The following factors should be taken into account [26.928]: • Tracking and Sensing: 3DoF tracking may require < 1 Watt, while 6DoF tracking requires more power consumption. • Display: display power consumption in the range of 0.3W to 1W. • Render (GPU): It can range from several mWatts to several Watts depending on resolution, frame rates and display technology. • Compute and Media Processing (CPU): similar to GPU, however, encoding may induce higher power consumption. • Connectivity: Wireless connectivity (for example connection to the network) power consumption can range from several mWatts to several Watts depending on bitrates, distance from radio access network, channel conditions, frequency range, etc. 4.2.7 Link to use-cases Reliable high data rate with bounded latency devices are the main enabling devices in fully merged cyber physical worlds use case: XR devices and it operates together with other body sensors/actuators (tactile gloves, electromyography (EMG) wristbands, smart watches, smart fabrics, etc.), and smartphone/connectivity puck. In a scenario involving fully integrated cyber-physical worlds, users will need to engage in communication with individuals located at a distance, aiming for interaction quality that closely resembles real-life experiences. Achieving this level of realism demands an improved understanding of body language, including gestures, intonation, facial expressions, surroundings, and sounds, among other aspects. Additionally, it involves enhancing other sensory perceptions, such as the ability to touch objects, enabling interaction with both physical and digital objects regardless of their proximity in the physical world. This kind of immersive experience use case is enabled using XR devices, as well as wearable devices such as earbuds, and devices integrated in clothes. 4.2.8 Example devices Figure 4-4: Different XR device types. Extended reality devices are of different types as shown in Figure 4-4. These types may differ in processing capabilities, communication types and power consumption. In AR, the display is usually transparent and digital information is superimposed onto real life objects. In VR, the display is not transparent and only virtual information and images are displayed in front of wearer's eyes. Also, body sensors/actuators (tactile gloves, EMG wristbands, smart watches, smart fabrics, etc.) can be integrated along with the XR devices. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 39 / 61 4.3 Highly reliable low latency device class The highly reliable low latency device class combines seamless connectivity, autonomous operation, and safety for human collaboration. With stringent latency demands, it excels in real-time decision-making. This class supports diverse communication needs, energy efficiency, and interoperability, ensuring seamless integration with various devices. 4.3.1 Key characteristics The key characteristics of this device class includes: Reliability: These devices are characterized by their unwavering reliability. They ensure seamless connectivity and efficient communication through dynamic resource allocation, guaranteeing uninterrupted data exchange. This attribute is paramount for applications that heavily depend on real-time interactions and responsiveness. Whether it's critical data transfer or swift decision-making, the reliability of these devices underpins their effectiveness. Defined by essential features for reliability and effectiveness, as outlined in [HEX23-D14], there is notable variation in reliability figures among different projects. For instance, the human and robot co-working use case demands a remarkable 99.999% reliability and minimal latency of 1 ms. Autonomy: Operating autonomously in diverse environments is a hallmark of this device class. Their ability to make informed decisions and adapt to ever-changing conditions while maintaining peak performance is remarkable. This autonomy empowers them to navigate complex tasks independently, making them invaluable in dynamic settings where adaptability is key. Safety: Safety is a paramount concern for these devices, especially when interacting and collaborating with humans. They are designed with an emphasis on safety to ensure cooperative and responsive engagement. This approach fosters an environment where these devices and humans can work harmoniously together, enabling shared tasks and collaborative efforts. Location Awareness: Another standout feature of these devices is their spatial awareness. They possess the ability to recognize their positions in relation to their surroundings, which enables them to adapt their behaviour based on their geographical context. This contextual awareness empowers them to make decisions and take actions that are not just timely but also relevant to the specific situation, thus enhancing their overall operational effectiveness. Low Latency: Due to their real-time operational nature, these devices have stringent latency requirements. For example, cobots, which are a prime example of this device class, demand ultra-low latency to enable realtime collaboration. The collaboration between humans and robots, exemplified in the 'human and robot coworking' use case [REI21-D11], tackles safety challenges, high-speed robot movements, and traffic volumes. The data rate is specified per user with high reliability and low latency (1 ms), aligning with Hexa-X's proposed 0.5-25 ms Roundtrip Time (RTT), while 1-50 ms for collaborating robots [HEX23-D14]. Notably, applications extend beyond cobots to include other devices like AGVs in this device class. In industrial settings AGVs rely on low latency to navigate and interact seamlessly within dynamic environments, enhancing overall operational efficiency. Versatile Connectivity: These devices exhibit versatile connectivity capabilities. They establish communication links not only with stationary objects but also with moving ones, such as AGVs even at speeds below 10 meters per second. This versatility makes them adaptable to a wide array of operational scenarios, ensuring effective communication and coordination. Multiple Frequency Bands: To overcome environmental constraints, such as line of sight blockages, these devices are designed to support multiple frequency bands. This flexibility is essential in maintaining robust and uninterrupted communication, particularly in scenarios where obstructions might disrupt signals. Innovative solutions like reflective intelligent surfaces can further enhance coverage at high frequencies. Energy Efficiency: These devices strike a balance between performance and sustainability by prioritizing energy efficiency. They are designed to minimize power usage, taking into account environmental concerns. This focus on energy efficiency not only ensures optimal performance but also reduces their overall environmental impact, aligning with broader sustainability goals. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 40 / 61 Interoperability and Standardization: Efforts towards interoperability and standardization are central to this device class. They are designed to seamlessly integrate with other devices, networks, and systems. This harmonization extends to aspects such as network architecture, data formats, and device interfaces, ensuring that these devices can collaborate effectively in a variety of environments and ecosystems. 4.3.2 Other characteristics Furthermore, there are various additional characteristics that might hold significant significance for the class of highly reliable low latency devices, depending on the specific demands of the given scenario. These characteristics encompass: Accurate positioning: It is essential for optimal resource management, safety response, asset tracking, etc. For instance, in the specific use-case of cobots, maintaining precise positioning (1 – 5 cm) must be supported at a 10 to 10000 Hz refresh rate. Accurate positioning is one of the crucial KPI as discussed in one of the usecase on smart transportation vehicles in [HEX23-D14] of achieving 1mm to 1cm every 20 ms. This precision is crucial not only for cobots but also for other devices such as AGVs, where real-time accurate positioning enhances navigation and coordination within industrial environments, ensuring efficient and safe operations. Time synchronization: It is crucial based on factors such as the specific use-case's network infrastructure, constraints related to latency, and the required levels of synchronization accuracy and precision to ensure seamless coordination and efficient data exchange. 3D positioning: It majorly holds significance in applications such as inventory management, enabling accurate tracking of items within a three-dimensional space. This capability enhances inventory accuracy, streamlines logistics, and optimizes resource allocation for improved operational efficiency. For instance, in devices like cobots and AGVs, where high reliability and low latency are paramount, 3D positioning plays a crucial role. These devices require precise spatial awareness to collaboratively navigate and interact with their surroundings. The combination of high reliability ensures consistent and error-free positioning, while low latency guarantees swift and real-time responsiveness, making them highly effective in dynamic and collaborative environments. Local computation: It encompasses on-device tasks such as data analysis, filtering, and inference, allowing devices to process information independently. This capability enhances efficiency by reducing the need for constant data transmission and central processing. Security and privacy: It is an essential safeguard for these devices to majorly ensure confidentiality and unauthorized access. Use-case specific: Example use-case, cobot’s environment mapping and object sensing with a range resolution <0.01 m and velocity resolution 0.1 m/s [6GB21-D21]. Other Features: This includes aspects like establishing safe axis range, detecting collisions, task scheduling and assignment, etc. 4.3.3 Technology enablers Key technology components will support the advancement of highly reliable, low latency devices, as discussed earlier. Among the various characteristics mentioned, this elaboration delves into the most promising enabler anticipated to uplift these devices as following: • Use of 6G spectrum for accurately sensing the space along with communication capabilities. By integrating radio-based sensing with cutting-edge sensor technologies such as LiDAR and 3D imaging, they enable devices to create detailed and reliable maps of their environment while avoiding obstacles. This integration promises breakthroughs in spatial awareness, benefiting applications like autonomous vehicles and smart cities. • Enhanced sensing with accurate perception and data collection, so as to enable devices to make informed decisions and respond effectively to changing conditions. Devices equipped with these advancements can perceive their surroundings with exceptional accuracy. As a result, they can make well-informed decisions and respond adeptly to dynamic conditions. This heightened operational intelligence has transformative potential across various industries, from healthcare to industrial Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 41 / 61 automation. In contexts where low latency and high reliability are paramount, such enhanced sensing capabilities enable devices to operate seamlessly, ensuring optimal performance in real-time applications. • Natural Language Processing (NLP) serves as a technological enabler, particularly in applications spanning applications like cobots, AGVs, etc. NLP empowers these devices to understand and respond to human commands swiftly and accurately, contributing to low-latency communication. By interpreting natural language inputs and gestures, NLP facilitates seamless and intuitive interaction between devices, ensuring effective collaboration. This technology significantly reduces the latency in communication, aligning with the stringent requirements of high-reliability applications. In scenarios like manufacturing, where precision and quick response times are crucial, integrating NLP enhances the overall efficiency and adaptability of the device class, making it out as a crucial facilitator. • Prioritize power saving techniques, secure communication, and effective energy management for reliability and safety. By optimizing energy consumption and implementing robust security measures, devices can maintain consistent performance while minimizing their environmental impact. Meticulous energy management further reinforces reliability, a crucial aspect for critical applications in fields like healthcare and transportation. 4.3.4 System/infra enablers Edge computing Enabling a highly reliable, low-latency device class necessitates a robust system and infrastructure, with a crucial focus on edge computing. These devices demand the capability to make split-second decisions while maintaining seamless connectivity and ensuring human safety. This involves equipping these devices with potent onboard processors, dedicated AI accelerators, and data pre-processing capabilities and then leveraging edge computing to complement this and significantly reduce latency, ensuring swift responses to real-time events. Additionally, edge computing enhances scalability, allowing the device class to adapt to varying workloads efficiently. While, onboard computing supports autonomous operation, enabling devices to execute tasks independently and adapt dynamically to their environment; the compute heavy tasks such as environment mapping can be done on the edge. It also ensures energy efficiency by optimizing the usage of computing resources, extending operational periods between recharges. Moreover, onboard computing enhances the device class's resilience, ensuring seamless operation even in environments with intermittent or limited network connectivity. RIS Another enabler that could support this class of devices is RIS. It could serve as an enabler, especially at higher frequencies (for e.g., sub-THz, mmW), to improve reliability and availability and enhance coverage. RIS may not be very relevant for <FR2 (as there are other alternatives such as multiple-input multiple-output (MIMO) to enhance coverage/ reliability). 4.3.5 Sustainability considerations The sustainability considerations for this device class are grouped under the following three sustainability pillars: Environmental sustainability: • Optimizing power consumption is a critical strategy for environmental sustainability. By optimizing power consumption, devices can significantly reduce their energy needs, leading to cost savings and a smaller environmental footprint. This approach aims to balance effective operation with minimal environmental impact, reducing the need for maintenance and ensuring long-term sustainability. • The maximization of renewable energy sources is essential for environmental sustainability. This involves integrating energy harvesting solutions into IIoT devices. These solutions, such as leveraging solar panels and wind turbines, enable devices to generate their energy, reducing their reliance on nonrenewable sources. This shift towards clean and sustainable energy aligns with global environmental objectives, contributing to an eco-friendlier future. Social sustainability: Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 48 / 61 techniques based in time, frequency, or power domain, without compromising the device’s overall energy usage targets. • Coexistence with 6G air interface and core network: Similarly, to LTE-M and NB-IoT evolutions for efficient coexistence with NR and support of connection to 5GC, the enhanced mMTC devices should have ensured coexistence with 6G air interface and support of 6G core network. In case of using legacy cellular IoT technologies, 6G air interface could be designed ab initio to support dynamic spectrum sharing (as was the case with NR), while enhancements may need to be provisioned in order to accommodate peculiarities of legacy technologies and device behaviour, e.g., always-on legacy signals. Additionally, support of connection to 6G core network should also include support of the various legacy radio resource control (RRC) modes and procedures, e.g., RRC idle/inactive, extended DRX, etc. 4.4.4 System/infra enablers mMTC type devices will need to be connected everywhere and anywhere, and certain cases will require a very large connection density. The necessary infrastructure and system design must be in place in order to provide the required connectivity and backend support for feasible functioning of IoT applications based on mMTC devices. The main infrastructure enablers envisioned include: • Ubiquitous low-power wide-area network deployment (base stations, gateways, management platforms) to support IoT applications suitable for mMTC devices requiring, e.g., close to 100% network availability or worldwide coverage, remote management of devices for configuration/monitoring at scale. • Network capability to support concurrently a large device connection density, with capacity for scalability, load balancing and resource orchestration, while also maintaining the per device necessary communication performance requirements (e.g., high redundancy, availability and reliability, accurate and timely location services, etc.). • Cloud and edge computing platforms to support potential needs of data storage, processing and analysis offloaded from mMTC devices. • Efficient integration of connectivity technology with network architecture and novel systems for mMTC services billing and subscription management. Such enabler seems crucial for targeting a device subscription cost of similar order of magnitude to the actual device cost and realising massive deployments in practice. 4.4.5 Sustainability considerations The sustainability considerations for the enhancement of mMTC device class are grouped under the following three sustainability pillars: • Environmental sustainability: o Low energy consumption and demand throughout the entire device life cycle, from manufacturing, to operation, to disposal. o Low resource and energy usage impact per device to the network. • Social sustainability: o Promote digital inclusion via ubiquitous and on-demand coverage (even for high connection density) o Promote well-being as well as digital inclusion by providing low-cost wearable and sensor services. o Promote social well-being, equity and security by providing for low-cost and ubiquitous emergency services for high connection density and preserving data privacy. • Economic sustainability: o Low-cost device solutions that can be deployed in mass scale. o Low resource and energy usage impact per device to the network. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 49 / 61 4.4.6 Key performance parameters In terms of radio performance, the following key targets are expected from mMTC type devices considering that they are targeting LPWA use cases and low-end massive IoT applications [May19]: • Connection density: up to 1 device per m2 • Coverage: up to 164 dB maximum coupling loss (MCL), at 160 bps data rate • E2E latency: few (10-15) ms to few (≤10) sec • Service availability: ≥99% • Data rate: from 10s kbps to few (<10) Mbps Considering that mMTC type devices will mainly refer to battery-based operation, energy consumption targets (considering the modem part of the device) should be supporting the use of relatively small size batteries for lifetime of several years, and should be similar to (or lower than) the power consumption of existing 4G/5G MTC solutions in the market: • Dormant state: ~1 µA • Reachable state: few 10s µA • Idle state: few mA • Active state: few 100s mA 4.4.7 Link to use-cases mMTC type devices are relevant to several use cases defined in [HEX223-D11] The most prominent ones include: • E-Health for All: Small, lightweight, low cost, and low maintenance mMTC devices and sensors, enabling for example on-body monitoring and basic e-health services that can be delivered anywhere, will enable accessible and remote healthcare. • Immersive Smart City: Resilient, low power/size/cost/maintenance mMTC devices and sensors for IoT, capable of coexisting under extremely high connectivity density numbers and in challenging coverage and interference environments, will enable management and risk prevention of city infrastructure through digital twinning. • Small Coverage, Low Power Micro-network in Networks for Production and Manufacturing: Large populations of sensors mounted in manufacturers’ machines and vehicles, mutually connected (possibly in a very trustworthy way) with very low-power, low-cost, small form factor and resilient mMTC devices, will enable energy and spectrum efficient production and manufacturing. 4.4.8 Example devices mMTC type devices include a large spectrum of terminals, ranging from sensors, meters and trackers to lowcost wearables and smart devices. Figure 4-6 summarizes some example devices in current market that lie within this mMTC type spectrum. Figure 4-6: Example mMTC type devices Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 50 / 61 5 Conclusion and Outlook Analyzing the use-cases in Hexa-X project such as [HEX21-D12], [HEX22-D13] and [HEX23-D14] along with the ones in Hexa-X-II [HEX223-D11], three distinct device classes that would be novel (compared to 5G device types) in 6G time frame were identified and a fourth one related to enhancement of existing 5G device type is also envisioned. Furthermore, from use-case and services trends that could manifest in 6G time frame, we could see possibilities for the following as future work: • Potential new device classes: o Enhancements of eMBB devices: This type of devices will primarily be about extending the data rates beyond 5G rates and making use of centimetric and sub-THz frequency bands in addition to frequencies supported in 5G. They will not have any mandatory requirements on latency, reliability, or availability. It is anticipated that one of the critical technology components that will enable this type of devices is that sub-THz frequency bands in addition to other frequencies such as mmW. Therefore, the current thoughts on the design methodology and scenario parameters related to sub-THz transceiver design is presented in Appendix A. o Enhancements of URLLC devices: Some literature, such as [ADS+22] points to insufficiency of current (5G) URLLC services that address 1ms latency at reliability of 1-10-5 (five 9s) for future services. This could motivate the need for enhancements of URLLC with lower than 1ms latency and higher than five 9s reliability along with higher data rates and massive connectivity. These challenging combinations are being motivated for the similar use-cases (or their extensions) cited in Section 2.1 and therefore motivate a through approach to understand first the deltas in the usecases and do a deeper analysis of the requirements. Hence this could be a potential device class by itself or an evolution of 5G URLLC device class. • Potentially novel device: Companion device: Analyzing CaaS and mainly to enable CaaS at extreme edge, there is a need for advanced & secure compute for low capability/resource constrained devices such as low-cost sensors. These additional computing needs could be served by another device, which acts as a companion to these resource constrained devices. Further investigations in conjunction with 6G use-cases details, new requirements (if any), system design is needed to conclude if it will be an independent device class. This is another candidate for future work while the current thoughts are presented in Appendix B. • JCAS and device classes: Joint communication and sensing (JCAS), wherein a wireless communication system is used to also provide the sensing capabilities, is seen as feature that finds application across different device classes and based on specific use-cases/scenarios. For example, in XR applications, it could be used to get a (near) real-time understanding of the environment that a user is traversing or in industrial scenarios where robots can detect a human nearby and activate safety features. The scenarios would determine the different parameters such as suitable spectrum (e.g., in sub-THz for higher resolution sensing), reference signal design, etc. in addition implementation choices such as shared signal processing blocks, special circuitry for Tx-to-RX cancellation etc. Thus, though there will be trade-offs based on the communication and sensing requirements, it is not seen as a separate device class, but rather a multi-mode device which could end up a sub-class with a device class. The Hexa-X-II use-cases work will continue as indicated in [HEX223-D11] analyzing further use-cases, with possibly understanding deltas with respect to 5G scenarios, understanding technologies that are needed etc. Thus, it is possible that some refinements of the device classes identified in this document and the potential new device classes could be continued in upcoming deliverables and along with analyzing the sensing aspects further. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 51 / 61 Figure 5-1: Summary of 6G device classes The above Figure 5-1 depicts the summary of device classes. It is good to note that the solid boxes indicate the identified 6G device classes, while dashed boxes depict the potential 6G device classes. The green filling and its gradient depict the requirements on energy consumption, with dark green representing strict requirements on energy consumption Though availability requirements are not explicitly shown, it follows the same direction as the reliability ones for the device classes. The following Table 5-1 contrasts the identified and plausible 6G device classes against the 5G ones. Table 5-1: An overview: 5G device classes against 6G device classes Legend 5G Device classes 6G Device classes Comments Novel • eMBB • URLLC • mMTC • Reliable High Data Rate with bounded latency (RHDRBL) • High Reliability and Low Latency (HRLL) • Energy Neutral (EN) • RHDRBL: high data rates with bounded latency to reliably serve for e.g., immersive experience use-cases. • HRLL: higher data rates than URLLC but less stringent reliability requirements targeting devices that have more safety requirements and autonomous operation including collaboration with humans. In the upcoming deliverables, this device class will be analyzed along with enhancements of URLLC together with developments in use-cases. • EN: enables energy efficient massively deployable devices. Enhancements from 5G - • Enhancements of mMTC (EmMTC) • EmMTC: improved cost and energy conservation, higher data rates while ensuring high coverage, possibly requirements on reliability, integrated Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 52 / 61 • Enhancements of eMBB (EeMBB) • Enhancements of URLLC (EURLLC) sensing capability, lightweight AI/ML techniques. • EeMBB: support higher data rates (and perhaps lower data rates than what is supported in 5G) but doesn’t have any latency requirements. It could be utilizing new spectrum bands in 6G. • EURLLC: potentially new spectrum (for e.g., in cm), physical layer enhancements, optimization enabled by ML/AI etc. could be interesting. Potentially novel from 5G - • Companion devices These devices mainly augment lower capable devices with secure, advanced compute and connect them via flexible topology. They may have only limited possibilities for deployments. More details in Appendix B. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 53 / 61 6 Appendix A: Sub-THz transceiver design Sub-THz bands could be utilized to achieve higher data rates given the wide bandwidths available. To realize communication over sub-THz bands, a critical component is a sub-THz transceiver that is studied in Task 5.2. The corresponding dimensioning approach and link to scenarios are described in this section. 6.1 Design methodology In order to design wireless systems, the following methodology is proposed and applied especially to sub-THz frequencies. Although the approach is relatively generic, the wider bandwidth and larger number of antennas associated to those frequencies bring specific design choices. A system is defined here as two or more transceivers communicating wirelessly with each other, and each transceiver contains the required components for physical layer (PHY) digital processing, analog front-end (FE) for baseband and RF, power amplifiers (PAs) generally considered separately from the rest of the FE, antennas with their interconnections, and device power supply. At least two transceivers are considered, one acting as transmitter and the other as receiver. In many cases a bi-directional link is created between them, primarily assuming TDD and swapping their transmitter/receiver role. This section focuses primarily on the downlink (DL) direction between one infrastructure transceiver (typically part of a base station) and a user equipment transceiver, but the approach is generic and can also include communication between identical devices. When more than two devices are considered, we assume that the infrastructure transceiver is a central node connected simultaneously to multiple user transceiver and offering them several parallel links. We do not consider more complex topologies such as multi-hop scenarios or network routing aspects. The objective is to dimension FE architectures and PAs (type of architecture, number of antennas, power levels, connections between components, semi-conductor technology, ...) and to specify the corresponding waveforms, signals and PHY digital processing algorithms. The main dimensioning and design steps are performed in the following order, with increasing complexity: • Start from scenario specifications. • Validate the target performance (mainly range and throughput) from link budget. • Assess and optimize the power consumption and energy efficiency of transceiver architectures. • Validate and optimize the physical layer algorithms under realistic conditions. In the next section, we list and shortly explain the specifications required in order to clarify scenarios, in such a way that the different steps of the methodology can be performed with sufficient information. The design flow is tuned towards mm-wave and sub-THz communications (mainly 30 to 300 GHz) but can be extended to other bands, with possibly a reduced accuracy due to less dedicated models of component performance, power consumption, or propagation. The methodology also primarily assumes zero-IF (direct conversion) architectures. The design flow mainly focuses on the in-band operation of the target system. Additional requirements may be included in order to also tackle interference aspects (from the target system towards other systems or vice versa). Based on the input scenario parameters specified in the next section, the proposed methodology determines the architecture offering the best solution by selecting and optimizing some of the following elements (not all of them are necessarily considered in each scenario): • Type of architecture (analog, digital, hybrid, ...) • Number of data streams, digital baseband streams, RF streams, PAs, antennas • Type of analog phase shifting • Specifications of DAC/ADC, phase noise, linearity, other non-idealities • PA specifications (output power, efficiency, technology) • Waveform, modulation and coding, frame structure for data and pilots • Tx and Rx PHY DSP algorithms at BS and UE side Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 54 / 61 6.2 Scenario parameters The relevant parameters to specify a scenario are listed in this section. For some of them, a unique value may be selected; for others, a range of possibilities or a list of requirements could be specified. Some parameters might come directly from the target application, while others could come from research interest, e.g., investigate a specific band or a specific category of architectures, or be specified by the system design. Practically, application requirements that should be specified directly from the application scenario are listed in 6.2.1, as they quantify the desired performance of the communication solution provided. Other more specific technical parameters partially come out of the system design effort and related constraints, models and investigations. They are listed in sub-sections 6.2.2 to 6.2.4. Not all input parameters are mandatory for each architecture investigation. Depending on the type of optimization desired, some may be irrelevant or left to default values. 6.2.1 Application requirements The end application requirements on the PHY connectivity should be expressed in parameters relevant to the wireless communication system design; some might be ignored in simple scenarios (such as single user): • Geometry o Target range between both transceivers (minimum, maximum) ▪ This may be generalized to include a distribution of users over the target range. o Angular spread of users as seen from the infrastructure transceiver o Minimal inter-user separation (in distance or angular) • PHY performance o Target active (peak) throughput per user (DL, UL) ▪ Possibly one minimal value and one desired (larger) value o Target BER (DL, UL) • Mobility of the users (maximum speed or distribution) • Number of simultaneously active users in the same band • Maximum application latency (limiting power-down phases when duty-cycling between two packets) 6.2.2 Band of operation and regulations In order to bound the design space, the main frequency and power parameters should be specified (they could reflect regulations, standards, or simple decisions on relevant scenarios): • Carrier frequency • RF bandwidth • Maximum EIRP (Equivalent Isotropic Radiated Power) per stream • Maximum total Tx power • Possible waveforms, constellations, coding rate and options • Maximum number of simultaneous users that can be spatially multiplexed. 6.2.3 Propagation channel Wireless propagation is a complex domain with many different environments and related models. Besides geometrical parameters above related to range and angular domain, a minimal and relatively generic description is proposed here; additional parameters may be needed depending on specific use cases. Some are illustrated on Figure 6-1. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 55 / 61 Figure 6-1: Tap-delay response where successive taps represent different propagation paths between the transmitter and the receiver, delayed based on the path length divided by the speed of light. • Type of propagation: Line Of Sight (LOS) or Non Line Of Sight (NLOS) • Delay domain statistics (see Figure 6-1) o Number of paths (the expected number of paths is the product of the density of paths by the maximum delay) o Density of paths (fraction of delay positions which are occupied by taps, at the channel sampling frequency which depends on the system bandwidth) o Maximum delay (corresponding to the last NLOS tap) o Tap amplitude distribution (the tap amplitude distributions are expected to lead a unit-energy normalized channel after integration of all taps over the delay dimension) • Additional losses for NLOS paths as compared to LOS o This parameter can be the adjustment variable such that the tap-delay energy is normalized to one in expectation, i.e., it represents the large-scale fading as compared to LOS propagation while the tap amplitude distribution represents the small-scale fading; in the LOS case this parameter is trivially 0 dB • Multi-antenna and multi-user correlations o Channel correlation between antennas of the infrastructure transceiver ▪ Options range from fully correlated (where elements of the same antenna array share the same channel except for phase shifts related to their position and path angles) to uncorrelated channels over the different antennas (mainly for low-frequency or distributed systems) o Channel correlation between antennas of a user transceiver ▪ The same model can be used at the user side if the antenna array is built with similar (antenna spacing) options. o Channel correlation between different users ▪ For multi-user scenarios, the correlation or independence of channels over multiple user positions is an important parameter that needs to be specified • Geometrical channel parameters o While the parameters above mainly refer to a statistical channel model combined with range and angles as specified in 6.2.1, another approach is to rely on a full geometrical channel model, e.g., using ray-tracing; in that case, other specific parameters are required to construct the channels: ▪ Room/environment geometry and dimensions ▪ Height and position of all transceivers ▪ Elements creating reflections/diffraction/attenuation and related material properties 6.2.4 Transceiver architectures Many parameters need to be selected or optimized in order to specify the design of transceiver architectures. This sub-section lists potential bounds or constraints that should be considered when selecting some parameters in order to obtain realistic architectures for the target scenario. It also includes specific sub-component Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 56 / 61 performance limitations that cannot easily be modified and hence are used as inputs for the analysis. Those values may be different for infrastructure and user transceivers: • Maximum consumed power • Maximum number of antennas and array size • Antenna element gain and interconnect loss. o Possibly antenna radiation diagram • Noise figure • Optimization criteria o Generally optimizing energy efficiency under performance constraints, but other options are possible. Hexa-X-II Deliverable D5.2 Dissemination level: Public Page 57 / 61 7 Appendix B: Companion devices In this Appendix, we outline a potentially a new device – companion device – that looks plausible in a 6G system and mainly acts as a node to augment computational needs of another device, especially resourceconstrained ones and possibly utilizing short range energy efficient links. Thus, these companion devices are mainly helping to realize CaaS (described in Chapter 2.2.1), which is then more of a use-case enabling service. The Figure 7-1 below depicts the companion device and its connectivity. Figure 7-1: Companion devices and network controlled flexible topologies. The companion devices are expected to interface with the served devices via a link under the control of the 6G system and accepts schedule from the 6G system. It should be able to simultaneously connect to both the Base station and its own served devices. The connections with served devices should be flexible, as in only if there is value in connecting, it should be set up. Thus, a flexible topology is needed among companion device and devices it serves, which can be set up by a 6G system that takes into consideration the authentication status, security requirements and resource capabilities of individual devices. Such flexible topologies also benefit from proximity detection of the companion devices and its potential served devices and is another feature that 6G system could provide. The availability and reliability of this topology could influence directly the QoS experienced by the end application. The connectivity (both between devices and between device & network) could utilize currently available spectrum and potential use the upcoming spectrum in centimetric range and sub-THz range. It is unlikely that sub-THz will be employed between device and network given the propagation path related constraints and connectivity will benefit from using licensed spectrum as interference could be reduced. It is to be noted that the flexible topologies with network control employed for CaaS could be one scenario that can be optimized for sub-THz transceivers (whose design aspects are covered in Appendix A). The companion devices would be different from traditional gateways in that it has dedicated functions that helps a remote device with its offloading functionality. Some examples of these functions could be: • Responds to mechanisms for discovery/detection of available compute resources (e.g., via a general register reachable by the CaaS provider). • Interfaces with a functional entity (e.g., central controller/workload orchestrator) that decides when to offload (fully or partly) a processing workload. • Ensures secure and trustworthy computation that doesn’t interfere with services offered to other served devices. This security features should also include the system-on-chip interfaces and benefit from trustworthy network-on-chip interconnects that help securely connecting to shared computation units.