Joint Device-to-Device Activation And Linear Transceiver Design In Wireless Cellular Networks

Direct communication between two or more proximal devices without the inter-vention of a base station(BS),known as device-to-device(D2D)communication,is a promising technique to improve performance of cellular networks in terms of spectrum utilization and overall throughput.It has been widely accepted that D2 D can provide proximity gain,reuse gain,hop gain,and pairing gain,etc.,and thus would be an important component in the next-generation(5G)wireless network.However,to achieve these potential benefits,D2 D communication must be appropriately managed.Some key issues in D2 D management include D2 D discovery,D2 D synchronization,mode selection/switch,wireless resource allocation,power control,interference miti-gation,and so on.In this paper,we concentrate on the mode selection and interference mitigation problem in D2 D communication,and then propose an algorithm based on joint mode selection and transceiver design in order to maximize the throughput of a multi-user multi-antenna cellular network.So far,various approaches have been proposed for D2 D mode selection and in-terference mitigation,covering the scenarios of single/multiple user(s),single/multiple antenna(s),perfect/imperfect channel state information(CSI),etc.Despite the differ-ences in problem formulation and algorithm design,most current studies divide the users into cellular users and D2 D users in advance,and fix their statuses during the entire communication process.Moreover,they usually take cellular users as primary users and thus their quality of service requests are delivered with priority,which is similar as the setting of primary users and secondary users in cognitive radio networks(CRNs).To guarantee this,cellular users are set to communicate with BS constantly,and only D2 D users are involved in mode selection.In this circumstance,one im-portant task of D2 D management is to limit the interference from D2 D users to cellu-lar users.For instance,to better protect cellular transmission from D2 D interference,D2 D users only reuse the cellular resources in uplink,while keeping idle in downlink.Obviously,this has significantly affected the flexibility of D2 D management.Thus,the performance gains of D2 D are not fully utilized.Actually,there are some key dif-ferences between D2 D and CRN.Specifically,the D2 D communication utilizes li-censed spectrum bands and is supervised by the cellular BS.Therefore,to maximally utilize the performance gains of D2 D,it may be more reasonable to assume that cel-lular users and D2 D users have the same priority.Motivated by this,we consider the underlay D2 D communication in multi-user multi-input multi-output(MIMO)networks in this paper.By jointly optimizing the transmission mode and the associated transceivers for each user pair,we aim to max-imize the network throughput.However,departing from current studies,we formulate this problem from a more general perspective.Specifically,we prioritize the cellular transmission and D2 D transmission equally.They share the same resources in both uplink and downlink.The transmission mode of each user pair can freely switch be-tween cellular mode and D2 D mode,depending on its contribution to the throughput and its interference to other user pairs.This setting relaxes the constraints on D2 D communication and formulates the D2 D communication problem from a more general perspective,thus helping maximally explore the D2 D potentials.However,this is a challenging NP-hard problem,and hence we seek some approximate solutions with manageable complexity.Our main contributions are summarized as follows.First,by performing the reformulation of weighted minimum mean square error(WMMSE),we design an algorithm which handles our problem via iteratively solving a series of WMMSE problems.Specifically,the WMMSE problem in each iteration is solved only inexactly to simplify the algorithm structure and accelerate the convergence.We show that the proposed algorithm is guaranteed to converge to a stationary solution.Second,we further fit the WMMSE problem into the framework of alternating direction method of multipliers(ADMM),so that the proposed algorithm can be executed in a distributed way.Finally,the problem can be solved distributively and efficiently,with a simple closed-form solution in each step.The efficacy and efficiency of the proposed algorithm are demonstrated by extensive numerical simulations.