Reconfigurable Intelligent Surface Assisted Mobile Edge Computing With Heterogeneous Learning Tasks

The ever-growing popularity and rapid development of artificial intelligence (AI) have raised rethinking on the evolution of wireless networks. Mobile edge computing (MEC) provides a natural platform for AI applications since has rich computation resources to train machine learning (ML) models, as well as low-latency access to the data generated by mobile and Internet of things (IoT) devices. In this article, we present an infrastructure to perform ML tasks at an MEC server with the assistance of a reconfigurable intelligent surface (RIS). In contrast to conventional communication systems where the principal criterions are to maximize the throughput, we aim at maximizing the learning performance. Specifically, we minimize the maximum learning error of all participating users by jointly optimizing transmit power of mobile users, beamforming vectors of the base station (BS), and the phase-shift matrix of the RIS. An alternating optimization (AO)-based framework is proposed to optimize the three terms iteratively, where a successive convex approximation (SCA)-based algorithm is developed to solve the power allocation problem, closed-form expressions are derived to solve the beamforming design problem, and an alternating direction method of multipliers (ADMM)-based algorithm is designed to efficiently solve the phase-shift matrix design problem. Simulation results demonstrate significant gains of deploying an RIS and validate the advantages of our proposed algorithms over various benchmarks. Lastly, a unified sensing-communication-learning platform is developed based on the CARLA simulator and the SECOND network, and a use case (3D object detection in autonomous driving) for the proposed scheme is demonstrated on the developed platform.

I. Introduction

The prevalence of mobile terminals and rapid growth of Internet of Things (IoT) technology have boosted a wide spectrum of new applications, many of which are computation-intensive and latency-critical, such as image recognition, mobile augmented reality, and edge machine intelligence. Mobile edge computing (MEC) is envisioned as a promising paradigm to ease the conflict between resource-hungry applications and resource-limited mobile devices, by providing cloud-computing capabilities within the radio access network in close proximity to mobile subscribers [1]. MEC is naturally well-suited for the AI-oriented networks, and the marriage of MEC and AI has given rise to a new research area, called “edge intelligence (EI)” or “edge AI” [2]–[5]. In general, there are two ways to realize the vision of edge AI, i.e., model sharing and data sharing [2], [6], [7]. Model sharing is typically achieved by federated learning which jointly exploits on-device training and federated aggregation, and a series of outstanding works focus on this type of edge learning [8]–[13]. However, running computation-intensive algorithms such as deep neural network models locally is very resource-demanding and requires high-end processors to be armed in the devices [2]. Moreover, training neural network models requires the data labels. In practice, however, the raw data collected by IoT devices, are generally unlabeled data and cannot be directly used for training. Therefore, we focus on data sharing where the data collected from the mobile devices (MDs) are offloaded to the MEC server for model training.