Adaptive Approaches for Blind Equalization Based on Multichannel Linear Prediction
Maurício Sol de Castro, João Marcos Travassos Romano · Anais do 2002 International Telecommunications Symposium · 2002
In this work, the problem of blind multichannel equalization is considered.A strategy for saving computation in zero-forcing equalizers design is proposed and its performance is evaluated under an adaptive implementation of an algorithm based on multichannel forward linear prediction.Moreover, a cascade structure based on forward and backward linear prediction is regarded: an adaptive implementation of such a structure is also proposed and its performance is verified through computer simulations. I. INTRODUCTIONNTERSYMBOL interference (ISI) is a major impairment in digital communications.Equalization is often considered as a suitable countermeasure for ISI.Usually, equalizer coefficients are adapted with a training sequence, which is required to be periodically sent.However, trained equalizers present important drawbacks such as wasted bandwidth and the possibility of fading occurrence during the training period.Blind equalization is then an interesting alternative, so that a training sequence is no longer needed.Blind algorithms that make use of higher-order statistics (HOS) are divided into explicit HOS-based algorithms and implicit HOS-based algorithms, which include the so-called Bussgang Algorithms.Both the implicit and the explicit HOS-based algorithms suffer from a slow convergence rate.Blind algorithms based on second-order statistics (SOS) are believed to overcome such a limitation.SOS-based algorithms exploit the cyclostationarity of the received signal.Such a property is preserved when the incoming signal is sampled at a rate higher than the symbol rate.It can be shown that such oversampling leads to a multichannel model.According to the Gardner's pioneer work [4], identification of both magnitude and phase of communication channels with SOS is possible due to the cyclostationary properties of modulated signals.Tong et al. proposed the use of cyclostationary SOS for blind channel identification and equalization [5].Most of recently SOS-based blind identification and equalization