Smart Solutions Towards Cellular-Connected Unmanned Aerial Vehicles System (AUTONOMY)
Lead Research Organisation:
Queen Mary University of London
Department Name: Sch of Electronic Eng & Computer Science
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Publications
Chen Z
(2024)
Adaptive Model Pruning for Communication and Computation Efficient Wireless Federated Learning
in IEEE Transactions on Wireless Communications
Chen Z
(2024)
Efficient Wireless Federated Learning With Partial Model Aggregation
in IEEE Transactions on Communications
Chen Z
(2024)
Robust Federated Learning for Unreliable and Resource-Limited Wireless Networks
in IEEE Transactions on Wireless Communications
Chen Z
(2024)
Adaptive Semi-Asynchronous Federated Learning over Wireless Networks
in IEEE Transactions on Communications
Chen Z
(2024)
Exploring Representativity in Device Scheduling for Wireless Federated Learning
in IEEE Transactions on Wireless Communications
Chen Z
(2023)
Knowledge-Aided Federated Learning for Energy-Limited Wireless Networks
in IEEE Transactions on Communications
Chen Z
(2024)
Distributed Digital Twin Migration in Multi-Tier Computing Systems
in IEEE Journal of Selected Topics in Signal Processing
Fayaz M
(2024)
Toward Autonomous Power Control in Semi-Grant-Free NOMA Systems: A Power Pool-Based Approach
in IEEE Transactions on Communications
Liu M
(2024)
A Nonorthogonal Uplink/Downlink IoT Solution for Next-Generation ISAC Systems
in IEEE Internet of Things Journal
| Description | The increasing demands for intelligent services, such as augmented reality/virtual reality (AR/VR) and Internet-of-Things (IoT) applications, motivate the integration of machine learning in future wireless networks. Federated learning (FL) is one of the most promising distributed learning frameworks to reduce the communication traffic load of intelligent services, which enables devices to collaboratively train machine learning models by periodically exchanging model parameters between devices and the parameter server instead of raw user data. The conventional model aggregation-based federated learning (FL) approach requires all local models to have the same architecture, which fails to support practical scenarios with heterogeneous local models. Moreover, the frequent model exchange is costly for resource-limited wireless networks since modern deep neural networks usually have over a million parameters. To tackle these challenges, we proposed a novel knowledge-aided FL (KFL) framework, which aggregates light high-level data features, namely knowledge, in the per-round learning process. This framework allows devices to design their machine-learning models independently and reduces the communication overhead in the training process. |
| Exploitation Route | Outcome of these findings could be used to develop intelligent transport system by the transport sector. Also, outcomes are useful for autonomous vehicle industries. It is also useful for healthcare to maintain the privacy of patients data. |
| Sectors | Digital/Communication/Information Technologies (including Software) Healthcare Transport |
| Description | Wireless Federated learning is one of the most promising distributed learning frameworks to reduce the communication traffic load of intelligent services in the next generation 6G wireless networks. Our findings in this project have been disseminated in numerous top IEEE journals and IEEE Flagship conferences and attracted the interest of non-academic industries. More specifically, we developed a novel knowledge-aided FL (KFL) framework, which aggregates light high-level data features, namely knowledge, in the per-round learning process. Our findings of KFL framework is highly useful for energy limited next generation wireless networks. This finding is highly useful and made non-academic industrial impact in the development of next generation secure 6G cellular networks. |
| First Year Of Impact | 2023 |
| Sector | Digital/Communication/Information Technologies (including Software),Healthcare,Transport |
| Impact Types | Economic |
| Description | Platform Driving The Ultimate Connectivity |
| Amount | £2,030,861 (GBP) |
| Funding ID | EP/X04047X/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 04/2023 |
| End | 04/2026 |
| Description | Scalable Hybrid Architecture for Wireless Collaborative Federated Learning (SHAFT) |
| Amount | £445,428 (GBP) |
| Funding ID | EP/W034786/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 03/2023 |
| End | 03/2026 |
| Description | TITAN Extension |
| Amount | £10,612,161 (GBP) |
| Funding ID | EP/Y037243/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 01/2024 |
| End | 03/2025 |
| Description | Collaborative Research on Federated Learning |
| Organisation | University of Cambridge |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | Proposed two mini projects for the TITAN Extension EPSRC Project (EP/Y037243/1), lead by University of Cambridge. |
| Collaborator Contribution | Two newly proposed mini projects are 1. AI-aided Pareto Optimization for UAV Swarm Trajectory Planning in NTNs Problem statement and brief state-of-the art summary (include an illustration if possible): multi-objective optimization (MOO) plays a pivotal role in the realm of airport flight planning, and its relevance extends to the domain of UAV Swarm Trajectory Planning in NTNs. In addition to conventional objectives, e.g., communication quality, energy consumption, and collision avoidance, UAV swarm introduces a new optimisation domain, i.e., swarm structures. During the planning phase, it is important to acknowledge the full potential of the considered system. The Pareto front (PF) becomes a pivotal indicator, with each point along this front representing an optimal solution. At the execution stage, we can choose the desired solution based on changeable constraints. Traditional MOO solution transforms the problem into a weighted single-objective optimisation problem, which only achieves one point on the PF. Some recent works use deep deterministic policy gradient (DDPG) algorithm to optimize this MOO problem, but the swarm structure optimisation and Pareto front are not included. Proposed new approach and methodology, provide a list of tasks and milestones: The first task is to design UAV swarm structures, where multi-agent reinforcement learning (MA-RL) can be developed to dynamically control the swarm by interacting with the environment. Based on the proposed swarm solution, the second task is to propose the PF via MO-soft actor and critic (SAC) algorithms. Milestones: 1) Design MA-RL algorithms for optimising UAV swarm structures; 2) Design MO-SAC algorithms for obtaining PF of the successful data packet rate and energy consumption under the fixed UAV swarm structure; and 3) Design multi-phase MO-SAC algorithms for obtaining PF for hybrid (centric + distributed) UAV swarm structure. 2. Federated Multi-modal Learning for Sensing and Communications Problem statement and brief state-of-the art summary (include an illustration if possible): Sensor fusion via wireless channels is important for sensing and communication systems as radiofrequency (RF) sensing has relatively low range resolution. However, different sensor belongs to different companies. The sensor fusion needs to consider both the multi-modal and privacy-protection problems. Existing works uses auto-encoder and federated average algorithms to solve this issue, but the auto-encoder structure design needs to manually access the local dataset on different agent and the federated average algorithm costs huge communication resources. Proposed new approach and methodology, provide a list of tasks and milestones: This project will first design a semi-supervised auto-encoder via variational Autoencoders to avoid leaking users' privacy. Then, this project will integrate the proposed auto-encoder algorithm into a new federated learning structure, where communication load via wireless channels will be minimized by considering the pruning and partial model transmitting. Milestones: 1) Semi-supervised auto-encoder algorithm for abstracting the common feature of multi-modal datasets; and 2) Federated multi-modal learning algorithm for the improve the performance of sensing and communications. |
| Impact | This collaboration started very recently. Research outcomes yet to come. |
| Start Year | 2023 |