Learning-Based Robust Resource Allocation for D2D Underlaying Cellular Network

In this paper, we study the resource allocation in D2D underlaying cellular network with uncertain channel state information (CSI). For satisfying minimum rate requirement for user and reliability user, attempt to maximize user’s throughput whilst ensuring a chance constraint D2D. Then, robust framework is proposed solving highly intractable constraint, where CSI uncertainties are represented as deterministic set enforced hold any within it. symmetrical-geometry-based learning approach developed model into polytope, ellipsoidal box. After that, under these uncertainty sets transformed computation convenient convex constraints. To overcome conservatism of approach, develop support vector clustering (SVC)-based compact set. Based on converted linear bisection search-based power algorithm obtained Finally, conduct simulation compare optimization approaches non-robust one.