With the rapid growth of the Internet of Things (IoT) and edge intelligence, federated learning (FL) has become a promising approach for distributed model training while preserving privacy. However, the resource constraints of IoT devices, including limited communication bandwidth and heterogeneous computational and energy capabilities, hinder the scalability and performance of FL. To address these challenges, we propose the Air-FedSNN framework, which integrates slimmable neural networks (SNNs) with over-the-air computation (AirComp) to improve communication, computation, and energy efficiency in heterogeneous IoT networks. A width-aware aggregation mechanism is developed to mitigate the impact of varying model widths and aggregation errors. Additionally, a dynamic width allocation strategy based on Lyapunov optimization and multi-armed bandit (MAB) theory is introduced to adaptively select model widths under long-term energy constraints, balancing energy efficiency and training performance without prior knowledge of device capabilities. Extensive experiments on MNIST, CIFAR-10 and CIFAR-100 datasets demonstrate that Air-FedSNN outperforms conventional FL methods, achieving superior accuracy-energy tradeoffs, and provides an efficient solution for resource-constrained environments.
Yue Zhang, Guopeng Zhang, Ke-Zhi Wang et al.· IEEE Journal on Selected Are...· 0 citations
Distinct from conventional integrated sensing and communication (ISAC) techniques, breakthroughs in LoRa-aided ISAC achieve hardware-unified sensing and communication capabilities for low-power devices. By combining such a novel technology with wireless power transfer (WPT), it yields wireless powered sensing and communication networks (WPSCNs). Information fusion, a widely adopted technique in such networks, relies heavily on the fresh fused information for effective system decision-making. However, age of information (AoI) is ineffective for measuring freshness of fused information. To tackle this dilemma, a novel metric, age of sensing (AoS), is introduced. Specifically, we study timeliness of a WPSCN, where a fusion center (FC) wirelessly powers sensing nodes (SNs) to collect sensing information from the SNs for generating fused information. Moreover, the impact of multi-cycle sensing on the AoS is first explored in the WPSCNs. We also adopt the adaptive transmission strategy for flexibly reducing the transmission duration. After obtaining a closed-form of the average AoS, it is then minimised by optimising WPT duration, multi-cycle sensing strategy and the SN locations. Ultimately, the numerical results validate the accuracy of our theoretical analysis. The effect of multi-cycle sensing and the superiority of adaptive transmission strategy are also demonstrated. Our findings offer valuable insights for analysing and improving the fusion system timeliness, and provide a theoretical foundation for the practical deployment of the WPSCNs.
Ya-Li Zheng, Shuai Shen, Ziye Xiang et al.· IEEE Transactions on Communi...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
