Blind recovery of M-QAM signals for fading channels using the fuzzy-c-means algorithm
Eugenio Costamagna, Lorenzo Favalli, Pietro Savazzi · 2003
In this work, the fuzzy-c-means (FCM) clustering algorithm is used to achieve joint equalization and demodulation of a QAM signal affected by intersymbol interference (ISI) in a blind mode, i.e. without insertion of a training sequence in the data stream. The points of the constellation are the cluster centers and are adapted to the current channel estimate iteratively by minimizing the FCM function. The clustering process takes place at every information packet and groups the received complex symbols by similarity without recovering an explicit estimate of the channel behavior. Results are presented showing both the BER improvement and the scatter plot of the equalized received signal.