Synchronization and channel estimation in DFT modulated filter bank cognitive radio transceivers

A discrete Fourier transform (DFT) modulated filter bank (FB) transceiver is investigated to be used for cognitive radio systems where the analysis FB at the receiver can be used for both sensing and spectrum access. A preamble design for the synthesis FB at the secondary user (SU) transmitter is presented which contains periodic symbols in analogy to short training fields in IEEE 802.11a. Based on the resulting frame structure a scheme for synchronization and channel estimation of the secondary user signal at the output of a delay-dispersive fading channel is investigated which does not require long training symbols. Here, symbol time offset (STO) and carrier frequency offset (CFO) estimates are initialized based on autocorrelation properties of the preamble signals. It is shown for a SU signal with a bandwidth of 30 MHz transmitted over an indoor radio channel that a subsequent space-alternating generalized expectation-maximization (SAGE) scheme with only three iteration cycles can be used to improve the STO and CFO estimates considerably and to provide additional estimates of the channel coefficients to be used for symbol detection.