Subchannel allocation with low computational and signaling complexity in 5g d2d networks

Abstract

Device-to-device (D2D) communication enables novel proximity services based applications in 5G networks. In underlay D2D, cellular users share subchannels with D2D users leading to interference between them. To efficiently manage the interference and assign D2D pairs to subchannels, we propose a novel relaxation-pruning algorithm (RPA). It allocates at most K D2D pairs per subchannel, where K is a system parameter that controls the trade-off between spatial reuse and inter-D2D interference. RPA is designed for a low signaling overhead scenario. In it, a D2D user feeds back a quantized rate to the base station that meets an outage probability constraint even though the user has only statistical knowledge of the inter-D2D and inter-cell interferences. RPA has polynomial-time complexity. It provably guarantees a D2D sum rate that is at least half of the optimal value, achieving which requires exponential complexity. This is unlike conventional approaches that offer no such performance guarantees or a weaker guarantee. Numerical results show that the D2D sum throughput of RPA is better than conventional algorithms and is within 1% of the optimal value.