Challenges and Directions for Deterministic Communication in 6G

Challenges and Directions for Deterministic Communication in 6G Gourav Prateek Sharma and James Gross Information Science and Engineering, EECS School KTH Royal Institute of Technology, Stockholm, Sweden Email: {gpsharma, jamesgr}@kth.se Keywords— 6G, URLLC, TSN, DetNet, wireless, ML, deterministic communications I. EXTENDED ABSTRACT Over the last decade, state-of-the-art in communications, networking and IT infrastructures has seen a steeply increasing interest in latency as - rediscovered - performance metric. In the wireless communications community, the pinnacle of this development has been the definition of Ultra-Reliable Low Latency Communications (URLLC), as one of three flavors of the 3GPP 5G mobile communications standard. In parallel, originating from tight latency requirements in Audio Video Bridging (AVB) applications, Time-Sensitive Networking (TSN) has been emerging as a major, unifying standard targeting Ethernet-based communications in vertical sectors such as industrial automation and manufacturing with stringent & deterministic latency constraints [1]–[5]. In a complementary initiative, the Internet Engineering Task Force (IETF) Deterministic Networking (DetNet) pursues similar goals but for L3-routed networks [6] and L4S for L3 queueing latency and congestion response [7]. Finally, with the advent of edge computing, the spatial proximity of public/open compute services translate into shortened access latencies, being one of the main advantages of edge computing over cloud computing [8]. These developments have to a large extent been triggered and motivated by use cases and stakeholders in the industrial automation and manufacturing vertical. Over the last 15 years, significant technical advances with respect to the design and operation of cyber-physical systems, robot systems as well as computer vision have incentivized the industry to renew digitalization efforts on the shopfloor, captured for instance in the concept of Industry 4.0 [9]. This, in turn, has enabled advanced manufacturing concepts like flexible automation, associated with a steadily increasing number of mobile systems on the shopfloor. 5G-based URLLC wireless communications, TSN, DetNet as well as edge computing systems cater to these needs. In addition, the automation community has been driving efforts to support shopfloor digitalization at scale through middleware standards like Open Platform Communications Unified Architecture (OPC UA). As of today, commercial systems based on all these standards have found their way already into industrial practice, mostly within small-scale scenarios embedded into controllable environments. Nevertheless, as digitization is continuously driving the transformation of society and industries, the natural progression is a substantial acceleration towards the 2030 timeframe, as more and more technical enablers mature and spark innovations. Along this trajectory, entirely new forms of interactions will appear, between people, between people and machines, and inbetween machines – all with the option of wireless communication. Furthermore, through cloud and edge computing, digital twins of physical entities and processes are enabled to create Cyber-Physical Systems (CPSs) that integrate physical and computational resources in order to control or monitor a mechanism. At the same time, artificial intelligence with data-driven system design opens novel dimensions for analytics and the optimizations of all sorts of processes. These trends are expected to result in a cyber-physical continuum, between the connected physical world of senses, actions, experiences, and its programmable digital representations as shown in Figure 1 [10]. This cyber-physical continuum is characterized by a massive scale-up and densification of CPSs such that increasingly these CPSs interact with each other at runtime. However, a prerequisite for this cyber-physical continuum is an intelligently networked infrastructure that enables scale and integration beyond controllable environments, forming an efficient “glue” between the digital and physical worlds [11]. In this talk, we argue that currently there is a unique window of opportunity to set the foundations of future 6G mobile networks to enable this emerging cyber-physical continuum. Two main goals need to be reached in this respect, going far beyond current standards and systems. On the one hand, scalability needs to be achieved among future infrastructures and CPS Figure 1 Illustration for the cyber-physical continuum. applications. On the other hand, convergence among standards and infrastructures towards an efficient provisioning of end-toend properties for CPS applications will have to be achieved. Let us start with a discussion around the related challenges with respect to the goal of convergence. Typical applications of the cyber-physical continuum will comprise CPSs such as control and automation, wearable robotics and exoskeletons, as well as Extended Reality (XR). All these CPS applications comprise of intimately coupled communication as well as compute loads, while requiring deterministic (i.e., guaranteed) “end-to-end” performance. Taking wearable robotics as an example, the interworking of Occupational Exoskeletons (OEs) allows reducing the physical load of a human worker which nevertheless imposes demanding requirements on computation and supplicated hardware [12]. Computational offloading, such as to the edge, offers significantly more efficient implementations of OEs in terms of energy consumption, efficiency and costs. However, such computational offloading necessitates a new type of support by the infrastructure in the “end-to-end” context. It is paramount here to stress the difference between the notion of “end-to-end" with respect to applications in the future cyber-physical continuum, versus its connotation traditionally in the design of networks. In the future, sufficient end-to-end performance refers to supporting the application communication and computation demands over the entire loop, from sensors via cyber-physical representation and back to actuation modules. This end-to-end loop might comprise different wireless and wired layer 2 technologies, incorporates either local or cloudified compute resources and thus spans over various different technologies, standards as well as administrative domains. End-to-end paths through such heterogeneous network infrastructures have today unpredictable latency, reliability and availability variations which directly jeopardize the operation of CPS applications. The main challenges relate therefore to an efficient integration of future cellular systems, TSN and DetNet domains, OPC UA as well as edge computing resources towards a deterministic end-to-end communication & compute infrastructure as shown in Figure 2. This will not only comprise Key Performance Indicators (KPIs) such as latency, reliability and availability, but also relates to time awareness, security, privacy and trust in the novel end-toend connotation applicable to CPSs. Scalability challenges of such future infrastructures are related to the ones regarding convergence, but significantly go beyond them. Beyond convergence of infrastructures, capturing CPS communication & computation loads, as well as their requirements and expressing their characteristic interaction patterns from an end-to-end perspective is paramount to have infrastructure adapt to CPS applications in an automated fashion. Corresponding interfaces and specifications are required. Likewise, converged end-to-end communication and compute infrastructures will have to adapt to a wide span of CPS applications in terms of their workloads and requirements, in order to enable scalability. Such an adaptation only becomes possible if converged end-to-end infrastructures become predictable in terms of their resource-KPI trade-offs while being subject to various stochastic influences. It is in particular the end-to-end relevance of future CPS applications that make the aspects of interfaces as well as predictability extremely challenging. These emerging challenges give an idea in which sense future 6G technology must ensure end-to-end deterministic characteristics across a multiplicity of heterogeneous technologies and application domains, including wired and wireless communication infrastructure as well as compute resources. Achieving this will require substantial extensions and harmonization among future 3GPP cellular systems, TSN, DetNet and other technologies/frameworks such as OPC UA. Beyond current standards and systems, state-of-the-art has partially addressed some of the above-mentioned challenges, however, to a large extent novel approaches are required. In this talk, we argue that deterministic communication systems are at the initial stage of their evolution rather than at their endpoint. Despite the intense research interest from academia and industry over the last decade, leading to initial technological standards and commercial products, it is clear that substantial further challenges need to be addressed to achieve scalability and convergence towards a cyber-physical continuum. Key concepts to overcome these challenges will be 6G-native evolution towards deterministic communication, wireless-friendly integration regarding TSN and DetNet, data-driven characterizations of stochastic elements, end-to-end time awareness as well as security concepts, as well as leveraging network digital twins towards system awareness. ACKNOWLEDGMENT This work was supported by the European Commission through the H2020 project DETERMINISTIC6G (Grant Agreement no. 101096504). REFERENCES [1] ‘IEEE Standard for Local and Metropolitan Area Networks–Timing and Synchronization for Time-Sensitive Applications’. Jun. 2020. Figure 2 Goals of the future deterministic system and the associated challenges. [2] ‘IEEE Standard for Local and metropolitan area networks – Bridges and Bridged Networks - Amendment 25: Enhancements for Scheduled Traffic’. Mar. 2016. [3] ‘IEEE Standard for Local and Metropolitan Area Networks - Virtual Bridged Local Area Networks Amendment 12: Forwarding and Queuing Enhancements for Time-Sensitive Streams’. Jan. 2010. [4] ‘IEEE Standard for Local and Metropolitan Area Networks–Bridges and Bridged Networks – Amendment 31: Stream Reservation Protocol (SRP) Enhancements and Performance Improvements’, IEEE Std 8021Qcc-2018 Amend. IEEE Std 8021Q-2018 Amend. IEEE Std 8021Qcp-2018, pp. 1–208, Oct. 2018, doi: 10.1109/IEEESTD.2018.8514112. [5] ‘IEEE Standard for Local and metropolitan area networks–Frame Replication and Elimination for Reliability’. Oct. 2017. [6] N. Finn, P. Thubert, B. Varga, and J. Farkas, ‘Deterministic Networking Architecture’, Internet Engineering Task Force, Request for Comments RFC 8655, Oct. 2019. doi: 10.17487/RFC8655. [7] B. Briscoe, K. De Schepper, M. Bagnulo, and G. White, ‘Low Latency, Low Loss, and Scalable Throughput (L4S) Internet Service: Architecture’, RFC Editor, RFC9330, Jan. 2023. doi: 10.17487/RFC9330. [8] W. Z. Khan, E. Ahmed, S. Hakak, I. Yaqoob, and A. Ahmed, ‘Edge computing: A survey’, Future Gener. Comput. Syst., vol. 97, pp. 219– 235, Aug. 2019, doi: 10.1016/j.future.2019.02.050. [9] K. Zhang, Y. Shi, S. Karnouskos, T. Sauter, H. Fang, and A. W. Colombo, ‘Advancements in Industrial Cyber-Physical Systems: An Overview and Perspectives’, IEEE Trans. Ind. Inform., vol. 19, no. 1, pp. 716–729, Jan. 2023, doi: 10.1109/TII.2022.3199481. [10] ‘6G – Connecting a cyber-physical world’. https://www.ericsson.com/en/reports-and-papers/white-papers/aresearch-outlook-towards-6g (accessed Dec. 11, 2022). [11] ‘Future network trends driving universal metaverse mobility’. https://www.ericsson.com/en/reports-and-papers/ericsson-technologyreview/articles/technology-trends-2022 (accessed Dec. 11, 2022). [12] ‘Occupational Exoskeletons: Overview of Their Benefits and Limitations in Preventing Work-Related Musculoskeletal Disorders’. https://www.tandfonline.com/doi/epdf/10.1080/24725838.2019.1638331 ?needAccess=true&role=button (accessed Dec. 11, 2022).