Latency Minimization in Intelligent Reflecting Surface Assisted D2D Offloading Systems

In this letter, we investigate an intelligent reflecting surface (IRS) aided device-to-device (D2D) offloading system, where IRS is employed to assist in computation from a group of users with intensive tasks another idle users. To minimize the system latency while cutting down heavy overhead exchange channel state information (CSI), study joint design beamforming and resource allocation on mixed timescales. Specifically, high-dimensional passive vector at updated frame-based manner based statistics, each frame consists number time slots, ratio user matching strategy are optimized relied low-dimensional real-time effective coefficients slot. A novel mixed-integer stochastic successive convex approximation (MISSCA) algorithm proposed tackle challenging problem. The convergence property computational complexity then examined. Simulation results show that our significantly outperforms conventional benchmarks.