Optimum and Suboptimum Blind Channel and Symbol Estimation for SISO Channels


Abstract: We present three methods for blind channel and symbol identification from a single or multi-block observation. These methods are deterministic approaches suitable for the identification of quickly changing wireless channels. The first method uses the finite alphabet property and it has good performance even for noisy observations. It requires only a single data frame, which is a unique feature of the method. This method can also be used to identify the channel order. For multi-block observations, we present the maximal ratio combining cross relation (MRCCR) method. It is an optimum approach in terms of instantaneous SNR and is based on the cross relation and maximal ratio combining techniques. The equal gain combining cross relation (EGCCR) is a suboptimum alternative to MRCCR with low computational complexity. MRCCR and EGCCR methods require only two frames for estimation and their performances are considerably better compared to alternative methods when the number of data blocks is small. In addition, they can perform as long as the data block length M >= 1. These three methods are especially well suited for trailing zero or burst transmission schemes.

Keywords: Blind channel identification, SISO, maximal ratio combining, trailing zeros

Full Text: PDF