A reliable method for blind channel identification using burst data

In this paper we present a new approach for blind identification of single input single output (SISO) and multiple inputs single output (MISO) FIR channels with nonminimum phase. The approach is based on minimizing a cost function built of the problem unknown parameters and a vector of measurements achieved by passing the received data through a known parallel set of FIR filters followed by samples averaging. Averaging is done according to certain functions that have higher order statistics (HOS) content and that their asymptotical mean can be expressed in closed form. The main advantages of this approach are its high probability of identification success when considering statistical channels, its ability to obtain reliable channel estimates in low SNR using short records of samples and its unsensitivity to overestimation of the channel order.