Blind Order Estimation Based on Subspace Identification and Weighted Least Squares Equalization

Channel order detection is significant for blind channel identification and equalization methods. Most existing order estimators are unable to work in practical microwave scenarios: low or moderate SNRs and ill-conditioned channels. In this paper, we propose a combined order estimation criterion by combining the extreme eigenvalue cost function and the weighted least squares equalization. This new criterion ensures the global minimum at the correct or effective channel order in noiseless case. For low or moderate SNRs, the proposed order estimation algorithm can still work with short samples and ill-conditioned channels. Simulation results demonstrate the superiority of this algorithm over other existing methods.