QoS-Aware Admission Control and Resource Allocation in Underlay Device-to-Device Spectrum-Sharing Networks

Device-to-device (D2D) communications underlaying a cellular infrastructure have been recognized as an important network-organization architecture in 5G networks. In these scenarios, existing works have explored the impacts of one or several factors among mode selection, admission control, partner assignment, and power allocation on the network performance. In this paper, we put forward an optimization framework that considers all of these coupled factors to investigate the spectrum sharing problem in D2D networks. In particular, we introduce an objective that combines the access rate and the network sum rate and then maximize it subject to users' quality-of-service requirements and resource allocation constraints. Due to its mixed combination, we focus on designing cost-efficient and easy-to-implement algorithms instead of finding globally optimal but exponentially complex solutions. By decomposition, we first devise two novel mode selection criteria and an admission-prioritized partner assignment scheme and obtain closed-form solutions for both admission control and power allocation. Moreover, we present a simple but interesting geometric interpretation on the physical implication of admission conditions. To further reduce the computational cost, we also exploit the structure of the formulation to devise a much faster heuristic algorithm, which usually runs at an order of millisecond. Simulation results show the low computational complexity of the proposed algorithms and exhibit their superiority against other schemes in terms of the access rate and the sum rate.