Optimal Semi-Persistent Uplink Scheduling Policy for Large-Scale Antenna Systems

In this paper, the uplink semi-persistent scheduling policy problem of minimizing network latency is considered for a training-based large-scale antenna system employing two simple linear receivers, a maximum ratio combiner and a zero-forcing receiver, while satisfying each user's reliability and latency constraints under an energy constraint. The network latency is defined as the air-time requested either to serve all users with a minimum quality-of-service, including reliability constraints and minimum throughput levels, or to maximize the spectral efficiency. Optimal non-orthogonal pilots are used to decrease the network latency. An optimization algorithm for determining the latency-optimal uplink scheduling policy using binary-integer programming (BIP) with an exponential-time complexity is proposed. In addition, it is proven that a linear programming relaxation of the BIP can provide an optimal solution with a polynomial-time complexity. Numerical simulations demonstrate that the proposed scheduling policy can provide several times lower network latency in realistic environments than conventional policies. The proposed optimal semi-persistent scheduling policy provides critical guidelines for designing 5G and future cellular systems, particularly for their ultra-reliable low-latency communication services.