Adaptive variable length Markov chain for non-stationary fading channel modeling

Abstract

A novel adaptive mapping from the measurements of a non-stationary wireless environment to a variable length Markov chain (VLMC) model is proposed in this research. This scheme consists of two main components: the estimation of channel signal-to-noise ratio (SNR) distribution and discrete VLMC modeling. To obtain the channel SNR distribution, a kernel density estimation algorithm is used to track local changes of channel statistics resulting from varying mobile environments. With the estimated channel SNR distribution, an iterative partitioning mechanism is performed to construct the VLMC model, which yields a much larger and structurally richer class of models than ordinary higher order Markov chains. The application of the derived VLMC channel model to the available throughput of a non-stationary wireless channel is examined with the feedback channel state information from the mobile terminal to the base station. The performance of the proposed adaptive mapping scheme and throughput estimation is demonstrated via simulation in a micro-cell non-stationary wireless environment.