Optimal Training Signals for Siso Ofdm Channel Estimation in the Absence/presence of Frequency Offsets

All existing training signal designs for channel estimation in OFDM systems assume no frequency offsets. In practice, frequency offset is unavoidable and seriously degrades the performance of OFDM systems. In this paper, we address the problem of designing optimal training signals for SISO OFDM channel estimation in the presence of frequency offsets. First, we present optimal training signals for OFDM channel estimation in the absence of frequency offset which include all existing optimal training signals for SISO OFDM channel estimation as a subset. Then we derive the optimal training signals for SISO OFDM channel estimation which are the most robust to frequency offsets. In the absence of frequency offsets, channel estimation mean square error depends only on the ratio of training signal energy to noise variance. In the presence of frequency offsets, channel estimation mean square error depends on this ratio as well as on the channel power delay profile and the frequency offset. Analytical and simulation results show that the performance improvement achieved by the proposed optimal training signals can be quite significant for moderate-to-high values of SNR and frequency offsets.