Adaptive channel identification and equalisation for GSM European digital mobile radio

It is proposed that maximum likelihood sequence estimation (MLSE) be implemented within the GSM European standard digital mobile radio system by use of the Viterbi algorithm (VA). In order for the VA to perform MLSE, it is necessary to supply the algorithm with an estimate of the impulse response of the transmission path. One way to achieve this is to utilize an adaptive channel identifier which must be trained periodically throughout transmission. In the proposed GSM Viterbi equalizers, the channel estimate is made at the center of the data block and then held constant throughout the entire message sequence. Due to the highly time-variant nature of the medium, the impulse response of the channel may, however, be subject to significant variation within one block period, resulting in estimation error and degraded equalizer performance. The ways in which identification inaccuracies affect the performance of the VA are considered. Theoretical expressions for expected channel impulse response (CIR) estimate error in the stationary and time-variant situations are derived. The use of adaptive algorithms to continuously estimate update throughout information transmission is also investigated. The theoretical performance bounds of these techniques are derived and validated by computer simulation