Resource Allocation for D2D Communications with Partial Channel State Information

Enhancement of system capacity is one of the objectives of the fifth generation (5G) networks in which deviceto-device (D2D) communications is anticipated to play a crucial role. This can be achieved by devising efficient resource allocation strategies for the D2D users. While most of the works in resource allocation assume full knowledge of the channel states, transmitting it in every time slot reduces the system throughput due to extra control overhead and leads to wastage of power. In this paper, we address the problem of D2D resource allocation with partial channel state information (CSI) at the base station (BS) and ensure that the interference from the D2D users do not jeopardize the communications of cellular users (CUs). With partial CSI, existing algorithms determine the Nash equilibrium in a distributed manner, whose inefficiency in maximizing the social utility is well known as the players try to maximize their own utilities. This is the first work in the D2D resource allocation field in which within a game theoretic framework, an optimal D2D resource allocation algorithm is proposed which maximizes the social utility of the D2D players such that a social optimum is attained. Each D2D player with the help of the BS learns to select the optimal action. We consider the channel to exhibit path loss. Next, we consider both a slow and fast fading with CU mobility and propose two heuristic algorithms. We validate the performance of our proposed algorithms through simulation results.