REDUCED COMPLEXITY FREQUENCY ESTIMATOR APPLIED TO BURST TRANSMISSION

In burst transmission, carrier recovery is a crucial point for synchronization systems. Indeed, an imprecise frequency estimation can cause cycle slips. An unbiased feedforward frequency estimator, for M-PSK modulated signals, in the non data aided mode (NDA) is derived from the maximum likelihood (ML) principle. A compromise is realized between noise filtering and estimation slip probability by minimizing the estimator variance. It is optimized to operate at low signal to noise ratio and short bursts. Its performance is compared to the ML estimator one. The estimator is applied in to a practical case (QPSK modulation in an all feedforward modem synchronization architecture in TDMA mode) with several burst lengths in order to exhibit the improvement due to estimation tracking with an integrating filter. The global performance of the modem synchronization unit is presented