Multiple hypothesis channel estimation for the MIMO-OFDM system

Abstract

This paper proposes a time-varying, Doppler channel estimation approach using multiple hypothesis channel modeling for OFDM-MIMO systems. A non-linear estimation structure is derived using this problem formulation which is composed of a parallel bank of extended Kalman filters whose individual outputs are weighted and summed to form channel estimates. By using this approach, we can reduce the model-mismatch errors associated with fixed-parameter channel models which can be a significant source of channel estimation error. For time-varying, Doppler channels, the channel models (e.g. ITU models) are characterized by a small set of parameters such as Doppler frequency and spectral shape.