Abstract : This paper investigates resource allocation for deviceto-device (D2D) communication in a millimeter wave (mmWave)cellular network where D2D users communicate in the cellularband. We formulate the optimization problem of subchannel andpower allocation, which aims to maximize the sum rate of D2Dtransmitters while satisfying interference constraints at the basestation (BS). Since the problem is NP-hard, solving it requires ahuge amount of computation. Therefore we reduce its computationalcomplexity by dividing it into two subproblems: a subchannelallocation problem and a power allocation problem. Tosolve the subchannel allocation problem, we propose two heuristicalgorithms: the max greedy SNR scheme and the minimum interferencescheme. The max greedy SNR scheme achieves highersum rate and has lower computational complexity compared to theminimum interference scheme. However it requires global channelstate information (CSI) of all nodes, which demands huge feedbackoverhead. On the other hand, the minimum interference schemerequires the location of each node. After subchannel allocation,the BS finds the optimal transmit power of each D2D transmitterby using the difference of convex (DC) programming. Simulationresults verify that our proposed subchannel allocation schemesoutperform the random subchannel allocation scheme, and the optimalpower allocation results in the improved sum rate of D2Dtransmitters.
Index terms : Capacitated max cut, device-to-device communication, difference of convex programming, hungarian algorithm, interference management, resource allocation, millimeter wave network, power allocation, subchannel allocation.