Novel recursive deconvolution noise compensation method for direct sequence radio channel sounding

Korhonen et al. (1995, 1996) showed that matched filter deconvolution (MFD), a form of minimum variance deconvolution (Chi and Mendel 1984), is a powerful inverse filtering technique for increasing time resolution in direct sequence (DS) radio channel sounding. A drawback of the technique is the impulse response estimate SNR floor due to deconvolution noise (DN). Cancellation of the DN is possible if spectral information at code zero frequencies can be regenerated (Korhonen and Haggman (1997). In this paper we suggest a recursive algorithm where the received spectra at code zeros is regenerated from the thresholded impulse response estimate. The algorithm is denoted as recursive spectral projection (RSP). We present a detailed flowchart, show some guidelines for parameter selection, and demonstrate the algorithm by simulation