Resource allocation for network-controlled device-to-device communications in LTE-Advanced

Network-controlled device-to-device (D2D) communication allows cellular users to communicate directly, i.e., without passing through the eNodeB, while the latter retains control over resource allocation. This allows the same time---frequency resources to be allocated to spatially separated D2D flows simultaneously, thus increasing the cell throughput. This paper presents a framework for: (1) selecting which communications should use the D2D mode, and when, and (2) allocating resources to D2D and non-D2D users, exploiting reuse for the former. We show that the two problems, although apparently similar, should be kept separate and solved at different timescales in order to avoid problems, such as excessive packet loss. We model both as optimization problems, and propose a heuristic solution to the second, which must be solved at millisecond timescales. Simulation results show that our framework is practically viable, it avoids the problem of packet losses, increases throughput and reduces delays.