The influence of Gaussian kernel width on indoor and outdoor radio channels identification from binary output measurements.

With the rapid evolution of digital communication systems and channels identification; there is a significant interest in the finite impulse response (FIR) filter theory, which has strong potential practical applications in various fields such as process control, signal processing, audio, and Hilbert transformers. In this paper, we are focused on the finite impulse response identification problem for single-input single-output nonlinear systems, whose outputs are detected by binary value sensors. In one hand, we have used the kernel recursive least squares (KRLS), and recursive least square (RLS) algorithms to identifying the practical frequency selective fading channels, called broadband radio access network (BRAN), standardised by the European Telecommunications Standards Institute (ETSI). In the other hand, the impact of Gaussian kernel width on the BRAN channels impulse responses identification and the mean square error (MSE) is also investigated. Monte Carlo simulation results in noisy environment and for various kernel sizes are presented to improve for what kernel size we obtain the optimal results.