(Semi) blind channel and data recovery in OFDM

Abstract

OFDM modulation combines the advantages of high achievable rates and relatively easy implementation. In this Thesis, we focus on the receiver design for OFDM transmission over block fading channels. Blind and semiblind techniques are developed for channel estimation and data recovery. The techniques use as many a priori information about the communication problem as possible to reduce the training overhead and/or improve the channel estimation accuracy and bit error rate. The thesis develops two blind techniques, a data centered technique (i.e. the input data is the main unknown) and the other is channel centered (i.e. channel is the main unknown). In the data centered blind technique, we show how an OFDM symbol can be blindly detected using output symbol and associated cyclic prefix. The approach is based on the transformation of the OFDM channel into two parallel subchannels due to the presence of cyclic prefix at the input. In the channel centered blind approach, the cyclic prefix and the Gaussian assumption on the transmitted data in an OFDM system are used for channel estimation and subsequent data detection. Finally, the thesis considers a more realistic problem in which we design a semiblind receiver for space time coded OFDM transmission over MIMO frequency selective, time-variant, and spatially correlated channels.