Study On Blind Multiuser Detection Algorithm In DS-CDMA Communication System Based On Signal Subspace

Code Division Multiple Access (CDMA) is an air interface standard of the third generation mobile communication system, whose excellent performance has been proved in all walks of life recently. However, we have to notice some problems brought by its disadvantages in the system. The spread spectrum codes of Direct Sequence CDMA (DS-CDMA) are not strictly orthogonal in a general way when mobile stations are connected stochastically, because the nonzero cross-correlation coefficient will cause multi-access interference (MAI). The interference not only limits the system capacity, but also brings Near-far effect, which severely degrades the performance when the number of mobile users increases or the signal power is enhanced. Therefore, MAI is regarded as the main restriction factor to the performance of CDMA system.Multiuser detection (MUD) combined with the smart antenna array is an efficient technique for interference suppression that can reduce the MAI, solve the Near-far problem and improve system capacity. Moreover, the blind MUD technologies are of great interest in the mitigation of MAI especially in the downlinks of CDMA systems, as only the prior knowledge of signature waveform and the timing of the desired user are needed without any training sequence and the interference users'information. Over the past decades, a significant amount of research has addressed various blind space-time multiuser detectors owning lower computational complexity and better performance.This paper mainly focuses on the blind multi-user detection algorithms based on the signal subspace in DS-CDMA system. At first, in this paper the basis of the MUD is introduced, which includes the basic principle and the basic structure of the MUD. Then the mathematical model of the blind MUD is built in the synchronous additive white Gaussian noise (AWGN) channel, and the vectors of blind decorrelating detector and minimum mean-square error (MMSE) detector based on signal subspace are given. The principle of blind MUD algorithms based on signal subspace is that the weight vectors of detector are expressed by signal subspace parameters.Through the simulations and studies on time-domain MUD with signal subspace, several algorithms with lower computational complexity and better performance are compared, such as classical subspace tracking algorithms PASTd and OPAST. We notice that these detectors with signal subspace estimation based on these classical tracking algorithms suffer from lower signal-to-noise ratio (SNR) that will considerably reduce their performance. The main reason is that the estimated signal subspace can not completely include the information of the desired signal under the conditions of lower SNR and insufficient data. Motivated by this observation, in this paper, a multiuser detector based on interference subspace (ISE) without loss of the desired information is presented. First, the interference subspace is estimated from the received data with the desired signal subtracted. Then, the signal subspace is spanned by the signature waveform of the desired user and the estimated interference subspace, which can fully contain the information of the desired signal. Thus, the proposed multi-user detector can offer better performance in lower SNR and with insufficient data. The simulation results show that the performance of the improved detector is significantly enhanced compared with that of the traditional ones, which could compensate for any performance loss caused by channel estimation and then make the bit-error-rate degrade.In this work, we also consider spatio-temproal blind adaptive multiuser detection schemes with antenna arrays. According to the simulation results, due to the restriction of the number of the spread spectrum codes used to differentiate different users, the capacity in time-domain MUD must be lower than N . Otherwise, there are some users using the same signature waveform, which can lead to the severe MAI. However, the capacity can be considerably increased to PN (P denotes the number of antenna arrays) by the implemention of antenna arrays, which is the most obvious and significant superiority. In order to alleviate MAI existed time-domain detectors, the expression of antenna array and the signal model in spatio-temporal MUD are introduced, and the vector optimization rule and spatio-temporal blind channel estimation are given.The other innovation in the paper is to combine the spatio-temporal MMSE detector with the interference subspace estimation, and the fast tracking algorithm (LORAF) and the classical tracking algorithms based on interference subspace in spatio-temporal detection are proposed. By the proposed approach, the desired signal steering vector can be resolved, which can be efficiently computed at a comparable cost with that of the classic method. Compared with the spatio-temporal version of the classical subspace-based MMSE MUD, the proposed approach exploits the spatial characteristic of users, increases the ability against MAI of users, and effectively extends the system capacity.