Minimizing Delay Violation Probability in URLLC over Fading Channels: A Cross-Layer Approach.

Yalei Wang,Wei Chen

DOI: https://doi.org/10.1109/globecom42002.2020.9348180

2020-01-01

Abstract:Ultra-Reliable and Low-Latency Communication (URLLC), aiming at meeting the hard delay constraint, plays a central role in 5G or even 6G mobile communications, thereby attracting much recent attention. However, satisfying such constraint over fading channels means a critical or even impossible mission for the physical layer because channel inversion exhibits a large power penalty in extreme fading environments. In this paper, we present a cross-layer approach to minimize the delay violation probability, the probability that hard delay constraint can not be assured, in URLLC over fading channels. More specifically, a probabilistic scheduling scheme that jointly exploits the queue state and the channel state is conceived. We formulate a Constrained Markov Decision Process (CDMP) to minimize the delay violation probability with a given power constraint. Variable substitution and linear programming are adopted to obtain the optimal solution. The optimal policy is shown to be threshold-based, which also induces the piecewise linear delay violation probability-power tradeoff. The tradeoff curve characterizes the theoretical lower bound of the delay violation probability with a given power constraint.