Non-Orthogonal Multiple Access for Delay-Sensitive Communications: A Cross-Layer Approach

Non-orthogonal multiple access (NOMA) has attracted great attention in the fifth-generation (5G) system to meet the rapidly increasing demand on quality of service (QoS). Among various QoS requirements, the urgent latency requirement has been expected to be provided for the delay-sensitive applications. In this paper, the delay-optimal uplink transmission in NOMA is studied over a block fading channel based on a cross-layer design, by which average latency is minimized with reliability provided by power allocation. In particular, the superposition coding in the physical layer and the scheduling in the network layer are jointly determined by the joint probabilities on the decisions of coding orders and transmission rates with the aware channel and queue states. With a constrained Markov decision process (CMDP), the cross-layer optimization is formulated to minimize the average delay subject to the constraints on power and reliability. The optimal delay-power tradeoff is obtained by formulating an equivalent linear programming (LP), which can be presented for multiple users based on a unified algorithm. Moreover, the optimal joint scheduling and superposition coding (JSSC) policy is constructed by using the structural properties. Based on the optimal cross-layer design, the transmission latency is optimized in the practical NOMA system.