Cross-layer modeling of AMC/ARQ-based wireless networks with outdated CSI

State-of-the-art wireless communications standards combine (hybrid) automatic repeat request (HARQ) protocols at the data link control (DLC) layer with adaptive modulation and coding (AMC) strategies at the physical layer. Most previous work on cross-layer modeling of these networks assumes the availability of ideal channel state information (CSI). However, in practical systems the CSI used to perform link adaptation may be outdated. This paper presents a generalized cross-layer multidimensional discrete-time Markov chain (DTMC)-based queuing model that jointly exploits the capabilities of ARQ and AMC, by explicitly accounting for imperfections in the AMC scheme caused by the delay in the CSI feedback channel. Based on the stationary state probability distribution of the proposed multidimensional DTMC, closed-form analytical expressions for performance metrics such as average throughput, queue length, packet delay and packet loss rate are derived, and an analysis of the impact of using outdated CSI on these performance parameters is presented.