Model selection-based blind channel estimation method for parsimonious MIMO-COFDM systems

The paper considers a blind channel estimation method combined with a model selection algorithm for the classification and estimation of significant channel parameters in coded orthogonal frequency division multiplexing (COFDM) systems with multiple receive antennas. Unlike conventional channel estimation methods, where channel parameters are estimated based on a known or an estimated length, we propose a blind-based method which will jointly classify and estimate only significant channel parameters and the channel length, i.e., parsimonious channel estimates. Firstly, by using only the received signal and userspsila spreading codes, a data model for the channel response is presented. Then, a blind channel estimation method based on eigenvalue value decomposition (EVD), incorporated with a hierarchical minimum description length (MDL) model selection method is used to estimate only the identified significant channel parameters. Simulation results show that, the proposed method is capable of identifying significant channel parameters with high probabilities and at low SNRs. In addition, the system performance based on the identified parameters is enhanced over conventional methods.