Optimal linear interpolative channel estimation for TDMA-based personal mobile communications over frequency-selective channels

Estimation of time-varying dispersive radio channels is one of the most important tasks of receiver synchronization. Second-generation TDMA-based mobile systems employ a training midamble for this purpose. It is proposed to replace the midamble by two training segments at both ends of an isolated time slot or packet. These are used to acquire two channel snapshots which, in a second stage, are linearly interpolated to yield channel estimates in the data segment of the packet. The interpolator weights are adjusted optimally, depending on the SNR in each acquired channel tap. A procedure for estimating the required tap SNRs from the two channel snapshots is presented also. The resulting channel estimator is shown by analysis and simulation to maintain near-optimal estimator performance over a wide range of unknown Doppler frequencies (vehicle speeds). The simulation results confirm the practicability and robustness of the new estimator, especially at high vehicle speeds