Study On Blind Multiuser Detection Algorithm In CDMA Communication System

IntroductionIt is evident that the 3rd Generation (3G) mobile communication will greatly facilitate and enrich our work and daily life, with the rapid development of modern digital mobile communication techniques. In order to provide colorful multimedia service and high rate data service, the 3G communication system need higher wireless capacity. The advantages of CDMA communication system are increased capacity, soft capacity, soft handover, higher spectral efficiency and the ability to resist the effects of multi-path fading, and so on. However, CDMA is self-interfered system, where multiple access interference (MAI) and Near-far effect are existing as the main factors restricting CDMA system capacity. At present, CDMA communication system mostly adopts power control technique to deal with MAI problem. The technique only mitigates MAI. The 3G mobile communication system regards multiuser detection (MUD) as an effective method to overcome MAI.MUD is an efficient technique for interference suppression that reduces the MAI and that solves the Near-far problem and improves system capacity. Traditional multiuser detector makes good use of all signals which result in multiple access interference so that it provides optimum MAI resistance. Nevertheless, there are some defects as follows: it assumes that the receiver can acquire the signature waveform and timing of desired user and the interfering users; it has no ability to suppress intercell multiple access interference; it can not be applied in downlink channels. Based on the traditional multiuser detection, adaptive multiuser detector eliminates the need to know the signature waveforms and the timing of the interferers and has to need training data sequences for every active user. But the frequent use of training sequence is certainly a waste of channel bandwidth. The blind adaptation schemes (BMUD) does not require training sequences and the interference user's information. It is an especially attractive research hotspot to the downlinks of CDMA lately because it only needs the prior knowledge of the signature waveform and the timing of the desired user, that is, the same knowledge as the conventional receiver.This thesis mainly focuses on the blind multiuser detection algorithm ofCDMA systems. In the paper the basis theory of multiuser detection and the existing blind multiuser detection algorithms are introduced, and the blind multiuser detection algorithms based on signal subspace, Kalman filtering algorithm are studied thoroughly. Aiming at the shortage of these methods, two algorithms of blind multiuser detection are proposed. One is the improved blind multiuser detection algorithm based on subspace tracking by introducing pairwise Gram-Schmidt orthogonalization, the proposed method can keep lower computational complexity, with faster convergence rates, better tracking ability and steady-state performance. The other is the improved kalman blind multiuser detection algorithm using the conception of subspace approach. It is shown that the detector can be expressed as an anchored signal in the signal subspace and the coefficients can be estimated by Kalman filter using only the signature waveform and the timing of the desired user, this proposed scheme has a lower computational complexity and better multiple path resistant.1. Blind multiuser detection algorithm based on subspace approachThe principle of subspace based blind multiuser detection algorithm is that the weight vectors of detector are expressed by signal subspace parameters, paper [15] first applied the subspace conception to the multiuser detection technology, proposed a subspace based structure of blind multiuser detection. The vectors of detector are calculated with adaptive subspace tracking algorithm, and the signal of the desired user can be obtained. When multiuser detector is adapted in blind mode, it usually adopts eigenvalue decomposition or singular value decomposition of received sample correlation matrix and tracking algorithms, which result in high computational complexity. At the same time, approximation computation in subspace tracking algorithms can also result in slow convergence.Through the simulations and studies on the subspace based multiuser detection, several algorithms with lower computational complexity and better performance are compared, and a better algorithm OPAST is given. After study on it, an improved blind multiuser detection algorithm based on subspace tracking is proposed to combat the divergence of the OPAST algorithm caused by numerical rounding errors. The proposed method introduces an efficient partial orthogonalization method, pairwise Gram-Schmidt orthogonalization, to mitigate this unsatisfied effect. Computer simulation experiments show that the improvedalgorithm has a much lower subspace tracking error, with global convergence and better steady-state performance. The new method has good stability and fast tracking ability even when SNR is low.2. Blind multiuser detection algorithm based on Kalman filtering The Kalman filter is known to be a linear minimum variance state estimator, and it is combined conventional kalman filter with adaptive algorithms. Zhang and Wei proposed a blind multiuser detection based on Kalman filter method in [56]. To develop the blind adaptive multiuser detector based on Kalman filtering, a novel linear first-order state-space model for the blind multiuser detection problem that is a perfect match for the application of Kalman filter theory in blind scenario is devised. The vectors of detector can be attained by calculating the minimum variance state estimation of Kalman filter state equation.The blind multiuser detection based on Kalman filtering has many merits. It shows better convergence compared with LMS methods, but the blind multiuser detection in multi-path fading channel is not studied and only synchronous or asynchronous situation is discussed in the paper. That is the performance degrades significantly in the presence of signature waveform mismatch. Aiming to this problem, the subspace technique is introduced and the detector is modeled as a vector in the signal subspace and employing a Kalman filter philosophy similar to that in [56] to derive the coefficients adaptively. Simulation results demonstrate that this proposed scheme has shown the advantages of lower computational complexity and faster convergence rate. In the case of signature waveform mismatch which could cause serious cancellation of the desired user's signal, it still works well while the Kalman detector suffers rapid performance degradation.