Channel estimation and blind equalization using minimum state per-survivor processing
A. Polydoros, Norman E. Lay · 2002
Per-survivor processing (PSP) is an adaptive form of sequence estimation which jointly estimates the data and the unknown parameters of a signal affected by memory such as a communications waveform corrupted by intersymbol interference (ISI). PSP maintains a corresponding channel estimate for each data sequence it retains and the Viterbi algorithm is used to select candidate sequences through the use of a standard trellis-based signal representation. Unfortunately, PSP exhibits an M/sup L-1/ processing complexity, where M is the size of the transmitted symbol set and L corresponds to the finite impulse response channel length. Minimum-state PSP ("mini-PSP") lowers the required computational burden for channel estimation by substituting a smaller value of M to create a "minimum state" receiver trellis. Data detection, is performed by inverting the channel estimate and linearly equalizing the received waveform prior to symbol by symbol decisions.