Interference Rejection and Channel Detection in Mobile Communication

Fredrik Nordström

Centre for Mathematical Sciences
Mathematical Statistics
Lund Institute of Technology

ISBN 91-628-5967-6

This thesis is devoted to statistical problems in mobile communication, in particular disturbance rejection and change detection are studied. Co-channel interference is a significant problem for systems based on TDMA, e.g. EDGE. We show methods for detection and suppression of such interference, when the terminal is equipped with only one antenna. Our algorithms are able to detect an interferer for practically relevant values of C/I and SNR. Further, we are able to estimate the interferering radio channel by correlation techniques and Kalman filters operating in a Viterbi decoder. After the channel is estimated the interference is suppressed by using joint Viterbi decoding or approximations of joint Viterbi decoding.
The radio channel for WCDMA system is subject to abrupt and fast changes, which are important to model and detect. We investigate two cases, disappearance of channel paths and appearance of new channel paths. The first case is solved by examining the tap power and alarm if it has been low during a long time. The algorithm is extended with another algorithm that uses the continuity property of a Rayleigh fading channel and locally approximates it by low order polynomials. For the second case three different methods are presented. The first one compares two different variance estimates. The second one is based on that the sum of thresholded residuals, within a window, is binomial distributed. The third method is based on the two sample Kolmogorov-Smirnov test. For both cases it is possible to quickly detect an abrupt path change, within 20-40 CPICH symbols. The mean time between two consecutive false alarms is designed to be the common time between two consecutive path searches.
 TDMA, EDGE, co-channel interference, Kalman filter, Viterbi decoder, WCDMA, detection of abrupt changes