Study On Blind Multiuser Detection Algorithms In DS_CDMA Communication System

Code Division Multiple Access (CDMA) has been choosed as the main way of multiple access in the third generation mobile communication system. The advantages of CDMA communication system are higher spectral efficiency, soft capacity, lower power, greater privacy, which have been reflected in the opened commercial CDMA communication systems. Yet the principle of CDMA technology is distincting all the users with their owned spreed spectrum sequence symbols, theories has proved that the binary spread spectrum sequences with both ideal autocorrelation and cross-correlation doesn't exist. It leads that the coefficients of cross-correlation which have been used to distinct users isn't zero. Non-zero coefficients of cross-correlation cause the interfere with each other of all the users,it calls multiple-access interference(MAI). MAI and near-far effect are the main factors that restrict CDMA system capacity and higher BER. At present, CDMA communication system mostly use power control technique to deal with MAI problem. The technique only mitigates MAI, but not removes it. With the third generation mobile communication is coming, the request of inhibiting MAI and near-far effect becomes more and more urgent to improve the performance of systems. Because Multiple-user joint detection can overcome MAI as an effective method in the actural systems, so more and more scholars begin to research it recently.The basic principle of MUD is fully using spread spectrum sequences, delays, and the corresponding signal processing methods to jointly detect all the users'informations in receiver, after that the desire user's information can be gotten. This technology can improve near-far effect wholly, lower the request of power control precision, use the spectrum resources more effectively, improve the capacity significantly. Traditional multi-user detector has some defects as follows: it assumes that the receiver can acquire spread spectrum sequence signature waveforms and timings of desired user and interfering users, when the numbers of user are huge, the algorithm frame can cause the computation very large, the capacity of the real-time treatment will become poor; adaptive multi-user detector eliminates the need to know the signature waveforms and the timing of the interferers, only to need training data sequences for every active user, but the frequent use of traing sequence is certainly a waste of channel bandwidth; the blind adaptation MUD doesn't require training sequence and the interference user's information, it only needs the prior knowledge of the signature waveform and the timing of the desirec user, that is the same as the conventional receiver. The effective method has attracted many scholars'attention recently and then all kind of blind detection algorithms have been proposed.This thesis firstly introduces the reasons of multiple access interference, the basic idea of MUD, the frameworks of the common linear and nonlinear multi-user detections,then mainly focuses on the blind multi-user detection algorithms of CDMA system based on tracking the signal subspace treatment and adaptive filtering,three directions are cocerned as follows:1. put forward an improved tracking signal subspace blind MUD algorithmFirstly introduces subspace information extraction theory of signal processing fields, then systematically descripts the technology of subspace blind detection, and the paper gives the subspace frameworks of tradisional detectors. The classic blind multi-user detection algotithm based on subspace need mathematical operations liked eigenvalue decomposition (EVD) or singular values decomposition (SVD), the corresponding operating time is very long, so the thesis mainly focuses on three tracking signal subspace algorithms that complexity is very low, these methods can achieve the real-time update of signal subspace. Improve the projection approximation subspace track and deflation (PASTd) with the schmidt orthogonalization algorithm, the improved method lowers the impact of weaking orthogonalization because of compression technology. Finally the simulation experimental proves that the improved tracking algorithms combines PASTd and OPAST's advantages, there are some raising along with near-far effect and bit error rate, especially the effect is obviously in resisting near-far effect when the interference powers are very high. The improved algorithm is suitable for the occasions which have bad power control, there is nor much advantage in resisting multipath fading,yet the computational complexity of improved algorithm increases significantly. When the users increase very fast in the actual CDMA systems, it should been considered to use carefully.2.introduce an improved LMS blind adaptive multi-user detection algorithmFirst this chapter descripts two models of the decision-making right rector and the constrained minimum output energy (CMOE) rules in linear blind MUD, then concentrates on two adaptive blind MUD algorithms: least mean square (LMS) and recursive least square (RLS) that follow the CMOE rules. Because the iterative factor impacts the traditional LMS algorithm convergence speed and precision very seriously, so this paper descripts a new LMS blind MUD algorithm based on decorrelation and variable step size, then compares the performance of above algorithms by simulink experiments. The result proves that improved LMS algorithm are better in convergence speed, the signal to interference and noise ratio (SINR) doesn't improve obviously.3.Apply the improved tracking signal algorithm to blind MUD based on Kalman filteringFirstly the thesis integrates the Kalman filtering algorithm in adaptive filtering field to blind MUD, reference the literature [ 48 ], a new linear first-order state-space mudel is proposed. It converts the solving problem of right vector into the minimum mean square error (MMSE) estimating problem that processes the parameters of Kalman filtering state equations, then compares the performance of Kalman, LMS, RLS blind algorithms by simulink experiments. Because in the environment of multipath fading the convergence precision and stability of the classic Kalman filtering algrithms decreases obviously, so this paper transforms the original state-space model with signal subspace processing, the improved algorithm change the detect model to a vector of signal subspace, the coefficients are gotten by estimating of Kalman filtering. The new Kalman filtering blind multi-user detection based on subspace use the above improved tranking algorithm. The simulink result shows that the convergence precision and stability of the improved Kalman filtering algorithm are better in multipath fading.