Kurtosis-based Blind Adaptive Detection Method

The work in this thesis is a part of the Nature Science Research Project of Anhui province "Research on blind adaptive multi-user detection in CDMA systems" (03042208) . In this thesis we study multi-user detection especially blind multi-user detection.At first, we make a summary and classify the existing blind multi-user detections till now.Secondly, we propose a multi-user detection based on kurtosis. This method can suppress not only MAI(Multiple-Access Interference)but also ISI(Inter-Symbol Interference). All needs in the detection are only the desire user's spreading code. It's a blind multi-user detection.Kurtosis is a kind of fourth order statistics and therefore has a high complexity. The signal's energy is made unitary by whiten before detection in order to simplify the expression of kurtosis. After that we also make use of the advantage of subspace separation to reduce the rank of the signal.Thirdly, In order to suppress the effect of time various channel, the adaptive signal processing techniques are studied and an improved adaptive method based on LMS (Least Mean Square) is proposed. The new adaptive method avoids the affection of amplitude and converges fast.At last we verify the performance of the improved adaptive method and the blind adaptive multi-user detection by simulations. It is shown that the improved adaptive method converges faster than LMS. The blind adaptive multi-user detection can suppress MAI and ISI at the same time, and its output BER performance is better than other blind or semi-blind detectors.