Fully distributed joint resource allocation in ultra‐dense D2D networks: a utility‐based learning approach

Power Allocation for D2D Communications Using Max-Min Message-Passing Algorithm

“…Moreover, the results of the proposed MP algorithm are compared with other works (e.g. [18], [19], [29], [38], [40], [41]) and centralized bench-marks such as exhaustive search results.…”

“…In addition, the achieved sumrate of D2D pairs with the rate of a single cellular user is depicted in Figs. 3(b) and 3(c) in comparison with two other existing game-theoretic approaches [18], [19]. The scheme proposed in [18] is a Stackelberg game designed for D2D sum-rate maximization.…”

“…3(b) and 3(c) in comparison with two other existing game-theoretic approaches [18], [19]. The scheme proposed in [18] is a Stackelberg game designed for D2D sum-rate maximization. Although the sum-rate of D2D pairs is maximized in [18], the share of the corresponding cellular user is low.…”

“…The scheme proposed in [18] is a Stackelberg game designed for D2D sum-rate maximization. Although the sum-rate of D2D pairs is maximized in [18], the share of the corresponding cellular user is low. Moreover, a hybrid-distributed scheme with the BS-supervision, proposed in [19], is depicted for two values of the maximum interference price denoted by θ max .…”

“…Their assumptions lead to an imperfect knowledge of network in realistic scenarios, which imposes a limitation on the convergence to Nash equilibrium, a reduction in precision of optimal solutions, and a delay in converging to equilibrium points. Moreover, the mentioned approaches also maximize the sum-rate of a cellular network and do not always result in [7] Power-channel allocation in D2D network Markov Decision process Explorative Q-learning Centralized Huge overhead [9], [18] Joint channel-power allocation Stackelberg game with pricing Stochastic learning algorithm Distributed Limitations to achieve Nash-equilibrium and the local optimality of achieved NE [11] Joint power control and mode selection A system throughput maximization problem subject to the QoS of users A distributed partly iterative algorithm Distributed The potentially high interference caused to the surrounding links is ignored. [12] Power allocation -A non-cooperative power control game Distributed with the BS control Sharing channel information and prices among users proposed a large signaling overhead [13] Power and mode selection A Bayesian Network (BN) A probabilistic approach Distributed using local information…”