Periodic Reporting for period 1 - RECOMBINE (Research Collaboration and Mobility for Beyond 5G Future Wireless Networks)
Période du rapport: 2020-01-01 au 2023-06-30
The capabilities and design dimensions of Future Wireless Networks (FWNs) or Beyond 5G networks are being driven by disruptive technologies such as artificial intelligence (AI), machine learning (ML), deep analytics, software and advanced computer technologies; advanced sensing, 3D imaging, virtual and augmented reality (VR and AR). This in turn drives the emergence of new applications that, however, are manifested by stringent performance requirements. FWNs will need to efficiently and flexibly provide diversified services such as enhanced mobile broadband access, ultra-reliable low-latency communications (URLLC), and massive machine-type (mMTC) communications. Based on the trends of hyper-connectivity, steadily emerging immersive applications and growing societal demand, FWNs will need to seamlessly unify the physical, digital and human worlds. This can be achieved through a new ecosystem of networks, sub-networks and device technologies, which will generate unprecedented economic opportunities. RECOMBINE addresses studies not only into the technology fundamentals, usages and experience design, but also the new business models that would allow for capturing the arisen social and market opportunities to meet adequately the Beyond 5G (B5G) communication networks vision projected for the year 2030.
WP1: T1.1 studied and designed novel algorithms and techniques in deep learning. Some initial research has been performed towards techniques to be applied for DSA/LSA for spectrum management. Contributions to T1.2 included the design and analysis of protocols for next generation green intelligent wireless networks by employing tools from game theory and stochastic geometry. Contributions to T1.3 were initial studies on novel business models for future wireless networks, with further contributions to be expected in M45. Several publications have been realised. WP2 contributed to the development of intelligence tools to support the MANO framework for FWNs; on measurements of radio signals, via drone-based receiver that records the in-phase/quadrature (IQ) radio frequency (RF) data, the development of a network slicing model, based on DLNN, and the development of an SDN-Controller for the implementation of custom functionalities such as KPI monitoring, network slicing and dynamic network resource allocation via a south bound interface. A basic version of the RIC was developed and later deployed on ICI lab’s private LTE/NR network servers. Further, an analysis of related work on semantic technologies for 5G and beyond networks was performed. Several publications have been realised. WP3 contributed to T3.1:Development of 3Dchannel models for UAV systems and UAV-based field measurements, T3.2: Development of antenna array beam-forming (BF) techniques using artificial intelligence T3.3: Design and optimisation of realistic antennas under multiple requirements; T3.4:Design of ultra-compact antennas suitable for UAVs, with a study and design of the initial antenna prototype for UAVs: i) Accuracy of 3D channel models for UAV systems & ii) Study and design of UAV - ground station wireless communications. Several publications have been realised. WP4-T4.1 research focused on proof-of-concept of a SLAM prototype using mm-wave frequencies; T4.2 on demonstration of a video distribution in a stadium; T4.3 Demonstration of security and safety design of UAVs (resulting in publication) and T4.4 Development of IoT algorithms for URLLC. WP5-Two deliverables have been submitted in support of the research. D5.1 Reference scenarios, system and user requirements and D5.2 KPIs and test cases. The work in this WP5 has been completed in M36 as per the workplan. Milestones: MS1, MS2, MS3, MS4, MS5, MS7, MS12 have been completed.
WP1 developed and evaluated AI techniques for spectrum sharing in next generation communication systems and researched use of LSA in future UAVs. Business models for using shared licenses for UAV industry have also been addressed. The potential uses cases of extending coverage in case of disaster and use as relays were identified and this work is continuing. WP2 contributed to the design of novel machine learning (ML) methods for data-driven network traffic characterisation and network status prediction and to advancing research in the enhancement of end-users’ perceived quality of experience (QoE). RECOMBINE advanced state of the art in AI algorithms for Network Prediction, Slicing, MANO, monitoring and QoE/QoS enhancement with both theoretical as implementation studies. Currently, work is continuing on the implementation of a RAN Intelligent Controller (RIC) inside the testbed located at the premises of STP, with the goal to ease the development of real-time AI algorithms. Future work envisions also embedding of the developed AI-based model for network traffic analysis and network slicing self-configuration in a real network environment, via the LTE/NR testbed deployed in ICI lab. WP3 performed UAV-based antenna and propagation measurements to enable low-latency communication links. Further, an improved antenna design was proposed as a baseline model to implement a novel technique to extend the low-frequency response. The use of neural networks (NNs) as a low-complexity beamforming technique has been proposed and has compared different structures of deep NNs in terms of accuracy and the temporal response while proposing a new beamformer implementation based on deep recurrent neural networks (RNNs), which is a novel area of research The requirements of use cases involving UAVs and also the main aspects needed for a realistic propagation channel models for future wireless networks have been investigated. Furthermore, the challenges related to the detection and positioning of UAVs in real-world scenarios have been studied and possible solutions to overcome these issues have been proposed.
WP4 contributed to advancing the research in energy-efficiency of user equipment through the development of AI-based algorithms for AOA and AOD. Also, contributions have been made to utilise mm-wave signals for simultaneous mapping and localisation (SLAM) and improving industrial processes. Future work will focus on further development and integration of the corresponding algorithms and the demonstration of their functionality in a proof of concept. Use of physical layer-based secret Key generation for generating secret keys by exploiting the wireless random channel followed by information reconciliation and a novel reconciliation approach based on a neural network have been proposed, which is an unexplored area. Work performed contributes to the use of URLLC services in the IoT context (Industry 4.0 connected vehicles, etc) and to the development of algorithms for dynamic re-configuration of network resources and device level parameters, which can maximise the overall performance of URLLC-type IoT applications. These algorithms exploit channel side information, as well as control information to trigger appropriate adaptation mechanisms and will be verified in future work on a proof of concept basis. WP5 defined reference scenarios and a set of key performance indicators (KPIs), which are linked to each scenario. A set of test cases for evaluation of RECOMBINE solutions have been defined.
WP4 contributed to advancing the research in energy-efficiency of user equipment through the development of AI-based algorithms for AOA and AOD. Also, contributions have been made to utilise mm-wave signals for simultaneous mapping and localisation (SLAM) and improving industrial processes. Future work will focus on further development and integration of the corresponding algorithms and the demonstration of their functionality in a proof of concept. Use of physical layer-based secret Key generation for generating secret keys by exploiting the wireless random channel followed by information reconciliation and a novel reconciliation approach based on a neural network have been proposed, which is an unexplored area. Work performed contributes to the use of URLLC services in the IoT context (Industry 4.0 connected vehicles, etc) and to the development of algorithms for dynamic re-configuration of network resources and device level parameters, which can maximise the overall performance of URLLC-type IoT applications. These algorithms exploit channel side information, as well as control information to trigger appropriate adaptation mechanisms and will be verified in future work on a proof of concept basis. WP5 defined reference scenarios and a set of key performance indicators (KPIs), which are linked to each scenario. A set of test cases for evaluation of RECOMBINE solutions have been defined.