RIS-assisted edge-D2D cooperative edge computing for industrial applications

Intelligence and low latency are particularly desired in the vision of industrial intelligence. To support intelligence, massive amount of data is generated from distributed Internet of Things (IoT) devices, and expected to quickly process with artificial intelligence (AI) for data value maximization. To reduce the network transmission delay from data source to computing nodes, edge computing is promising. However, the scarce and dynamic wireless resource as well as limited computing capability of edge servers challenge the edge intelligence. This paper studies a reconfigurable intelligent surface (RIS)-assisted edge-device-to-device (D2D) cooperative edge computing for end-to-end delay reduction in industrial environments. In special, we consider such an edge-D2D cooperative edge computing system, where the IoT devices could offload their tasks to edge server via cellular links for edge computing, or transmit the tasks to helper nodes via D2D links for local computing. We also consider the base station (BS)-based and D2D-based RISs deployed in cellular and D2D networks respectively. We formulate the joint computation offloading, beamforming optimization of both BS-based and D2D based RISs, and CPU resource allocation problem for average end-to-end delay minimization. Then, a distributed and cooperative scheme, called RIS-assisted edge-D2D cooperative computation offloading (RIS-assisted EDCO), is proposed to address the problem. The simulation results have illustrated the efficiency of the proposal for low end-to-end delay performance provisioning.