Completely Distributed Power Allocation using Deep Neural Network for Device to Device communication Underlaying LTE.

Device to device (D2D) communication underlaying LTE can be used todistribute traffic loads of eNBs. However, a conventional D2D link iscontrolled by an eNB, and it still remains burdens to the eNB. We propose acompletely distributed power allocation method for D2D communicationunderlaying LTE using deep learning. In the proposed scheme, a D2D transmittercan decide the transmit power without any help from other nodes, such as an eNBor another D2D device. Also, the power set, which is delivered from each D2Dnode independently, can optimize the overall cell throughput. We suggest adistirbuted deep learning architecture in which the devices are trained as agroup, but operate independently. The deep learning can optimize total cellthroughput while keeping constraints such as interference to eNB. The proposedscheme, which is implemented model using Tensorflow, can provide samethroughput with the conventional method even it operates completely ondistributed manner.