Analysis of backlog and delay in downlink power-domain non-orthogonal multiple access wireless networks

Abstract As important quality-of-service (QoS) performance indicators, backlog and delay in wireless networks must be in compliance with the requirement of low-latency and high-reliability scenarios. Non-orthogonal multiple access (NOMA) is a novel and key wireless technology for the next generation of mobile communications. Its potential impacts on backlog and delay deserve to be specially studied. Meanwhile, the stochastic network calculus is an effective tool for analyzing network performances such as backlog and delay. In this paper, for both orthogonal multiple access (OMA) scheme and NOMA scheme, we utilize the stochastic network calculus to characterize respectively cumulative service processes of channels in a downlink network. Based on moment generating functions, the closed-form expressions of the upper bounds on backlog and delay are derived through the definite integral and special function. Simulation results validate reasonableness and effectiveness of the proposed approach. We show that in terms of backlog and delay, performances cannot be always improved by substituting NOMA scheme for OMA scheme, except when the channel gain gap between users is significantly big. Moreover, the sensitivity to the same factor such as arrival rate, transmission power and power allocation, differs from user to user in NOMA. By simulations, it is further demonstrated that the performance loss due to a finite buffer in practical scenarios is acceptable and sometimes negligible, as long as the buffer size is set to be just moderate.