Performance Analysis of Cross-Layer Scheduling with Continuous States: A Quantized Markov Model

Lei Huang,Wei Chen

DOI: https://doi.org/10.1109/globecom54140.2023.10436924

2023-01-01

Abstract:Cross-layer scheduling is expected to play a vital role in low-latency or even real-time wireless communications with extensive applications in smart grids, automatic driving, and factory automation, etc. However, the performance analysis of joint channel and queue aware scheduling over fading channels with continuous channel states, e.g., Rayleigh channels, remains open. In this paper, we present a unified framework for performance analysis of cross-layer scheduling, which is capable of revealing the probability distribution of its queueing delays by formulating its approximate Markov models in low complexity through quantization. In particular, enlightened by the theory of stochastic order relations of random variables, we conceive two discretized Markov models, the queueing lengths of which are stochastically larger and smaller than that of the original scheduling policy with continuous states and decisions. In other words, the two quantized Markov models allow the upper and lower approximations, respectively, of a joint channel and queue aware scheduling. Moreover, we show that their queue length distributions converge fast as the quantization error vanishes. As a result, the quantized Markov model provides a unified framework for theoretical performance evaluation of cross-layer scheduling with continuous states and decisions. Simulation results demonstrate that the approximation error is small even in very low computational complexity.