Global Optimal Resource Allocation for Efficient FD-D2D Enabled Cellular Network

Abstract

This paper tackles the resource allocation (RA) problem of a full-duplex (FD) device-to-device (D2D) communications enabled cellular network. In the considered scenario, multiple FD-D2D pairs share the uplink channels of the regular cellular users (CUs) which leads to mutual interference between the two communication types. Within this interference environment, this paper aims to properly allocate the network's resources, such as the transmit power and the channels, to maximize a network-centric metric like the weighted-sum rate (WSR) and the global energy efficiency (GEE). The complex coupling between the mutual interference of the different links, as well as the flexibility of assigning the channels to the users, results in a non-convex RA optimization problem, for which the global optimal solution is hard to obtain. This paper is a first and innovative approach that globally solves the RA problem of an FD-D2D-based cellular network. In particular, we show that the global optimal solution can be achieved by decoupling the original problem into two sub-problems as power allocation (PA) and channel assignment (CA). The PA sub-problem is solved by means of monotonic optimization theory. Precisely, we propose a new polyblock-based algorithm, MARIO, which efficiently converges to the global solution of the PA problem. Then, based on the optimal PA solution, the CA problem reduces to an assignment problem, which can be solved by Khun-Munkers algorithm. Further, we propose a sub-optimal solution by solving the original RA problem in the reverse order, i.e., first assigning the channel and then allocating the power. The simulation results show the effectiveness of the proposed algorithms and provide important insights on the solution design parameters such as the proximity distance and the self-interference cancellation capability.

Game-Theoretic Based Power Allocation for a Full Duplex D2D Network

“…In this section, numerical simulations under different operating conditions are used to analyze the performance of the proposed distributed PA (Algorithm 1). Moreover, since none of the previous work address the distributed implementation of PA in the context of FD-D2D network, we compare Algorithm 1 with two power allocation strategies: i) the centralized power allocation algorithm in [13] which aims to maximize the global energy efficiency of the FD-D2D network (GEE) defined as GEE = k∈K R k k∈K p k +3Pcir , ii) the fixed power allocation strategy in which each user transmits at its maximum allowed transmission power P k max . Besides, three metrics are used to assess the performance: (i) the D2D rate gain brought by the accessed FD-D2D pairs defined as R gain…”

“…After that, we mathematically proved that this non-cooperative game has a unique equilibrium point that can be achieved through an iterative distributed process. The effectiveness of our distributed algorithm is verified through a numerical simulation which employs the centralized PA scheme proposed in our previous work [13] as a benchmark.…”

“…Thus, the works in [11], [12] proposed a PA scheme that maximizes the global energy efficiency (GEE) of an FD-D2D based cellular network. Using the Monotonic optimization theory, the authors of [13] derived the optimal PA scheme which maximizes the GEE of an FD-D2D network while fulfilling the QoS of the users.…”

“…To the best of the authors knowledge, none of the existing works provides a decentralized PA scheme of an FD-D2D network. To validate the proposed distributed PA algorithm a numerical simulation is conducted which employs the centralized PA scheme proposed in [13] as a benchmark. • Highlighting the impact of the SI cancellation factor and the D2D proximity distance on the performance of the FD-D2D network.…”

“…In this section, we aim at providing a distributed power allocation algorithm that converges to the unique NE point of the game G. To do so, we need to derive the point p k defined in (13). For ease of notation, we denote by µ k I −k +N 0 the equivalent interference gain on the k-th link, and call γ c max g c,bs , γ d1 max g d , γ d2 max g d the maximum SINRs values that can be obtained at the cellular link and the D2D links respectively when these nodes does not receiving interference and the thermal noise is negligible.…”