Resource Allocation for Latency-Aware Federated Learning in Industrial Internet of Things

Abstract

Federated Learning (FL) has been employed for tremendous privacy-sensitive applications, where distributed devices collaboratively train a global model. In Industrial Internetof-Things (IIoT) systems, training latency is the key performance metric as the automated manufacture usually requires timely processing. The existing works increase the number of effective devices to accelerate the training. However, devices in IIoT systems are usually densely deployed, increasing the number of clients can potentially cause serious interference and prolonged training latency. In this paper, we propose RaFed, a resource allocation scheme for FL. We formulate the problem of reducing training latency as an optimization problem, which is proved to be NP-hard. We propose a heuristic to select appropriate devices to achieve a good trade-off between the interference and convergence time. We conduct experiments using an RGB-D dataset in an IIoT system. The results show that compared to the state-of-the-art works, RaFed significantly reduces the latency by 29.9%.

Federated Learning Over the Industrial Internet of Things: A Joint Optimization of Edge Association and Resource Allocation

“…Therefore, it is of necessary to jointly consider the device selection and resource allocation for further optimizing the FL performance while meeting the delay and energy overhead requirements. Some recent works have focused on the joint device selection and resource allocation in FL-enabled IIoT systems [7], [20]- [22]. A stochastic optimization problem was developed in [7], aiming to minimize the FL evaluating loss by jointly optimizing the device selection, spectrum assignment, and computing resource allocation.…”

“…The authors in [21] adopted the DDPG to select IIoT devices with high data quality by minimizing the local training cost. In [22], the authors explored the problem of reducing the training delay by properly allocating the active IIoT devices and radio resources.…”

“…The aforementioned works [7], [17]- [22] have laid a solid foundation on resource allocation or joint device selection and resource allocation in the hierarchical FL architectures over IIoT systems. To our best knowledge, the issue of optimization in the utilization of multiple performance metrics, e.g., the learning speed and the energy consumption, in IIoT systems has not been addressed and remains an appealing study.…”

Enhancing URLLC in Integrated Aerial Terrestrial Networks: Design Insights and Performance Trade-offs

“…The advancement in the wireless communication industry is grown significantly over the last few decades [1]. It is expected that mobile data traffic will increase by 1000 folds to accommodate the internet of things (IoT) traffic [2]. To meet the dramatic increase in user demands, ultra-reliable lowlatency communication (URLLC) is considered one of the key applications for next-generation wireless networks that is more intriguing and challenging as it forces the quality of services (QoS) to achieve a delay of less than 1 millisecond and reliability greater than 99.99% [3].…”

Efficient Pruning-Split LSTM Machine Learning Algorithm for Terrestrial-Satellite Edge Network

“…The sixth-generation (6G) wireless network is anticipated to provide high data rate, low latency, low communication cost, and ubiquitous service [1]. This allows for the rapid development and popularity of the Internet of Things (IoT) [2]. However, traditional terrestrial networks are often difficult to fully cover remote regions or disaster areas to meet the demands for "ubiquitous service".…”