The Research On Multiuser Detection Technology In DS-CDMA Communication System

The multiuser detection technology is one of the key techniques of the third generation communication system. Its main thought is making full use of the information of all users to do the consociation detection towards the receiving signal ,as well as repress the multiple access interferences , alleviate the "far- near effect" and sequentially raise the system capacity availably.The main work of this paper is to carry on the research and analysis to the various multiuser detection algorithms within the DS-CDMA system. Further more, this paper puts forward both a blind adaptive multiuser detector based on subspace decomposition kalman filter and a blind PARAFAC multiuser detector.The optimal multiuser detector is put forward in 1986 by Verdu.S. This optimal multiuser detector can acquire the theoretically minimum bit error rate, providing the best detection performance. The optimal multiuser detector shows the potential in capacity and performance of CDMA system, it's the ideal appearance which we study the multiuser detection technology to pursue .At the same time, the optimal multiuser detector needs almost all user informations and its calculation complications growth exponentialy with the increment of the user count. when there are a great many of users ,it almost can't carry out.The abuse of the optimal multiuser detector make it only have theoretically meaning and can't to come ture. The decorrelation multiuser detector is put forward in1989 by R.Lupas etc.The multiple access interferences is caused by the out of vertical of the spread spectrum code waveform of different users. The main thought of the decorrelation multiuser detector is reliefing the multiple access interferences by relieve the correlation of the spread spectrum code waveform of all users to make the spread spectrum code waveform vertical. There are mang advantages of the decorrelation multiuser detector,such as having the best capacity to repel the " far- near effect" and having the ability to relief the multiple access interferences caused by the out of vertical of the spread spectrum code waveform absolutely,. Although the advantage of the decorrelation multiuser detector, there are still several disadvantages as it will enlarge the noise. The linear MMSE multiuser detector regard multiuser detection as a linear estimation problem. It can minimize the minimum mean squares error between the actually send out signal and the decision output signal throughout linear transform for it take into account the background noise and made use of the knowledge of the send out signal power. Thus, the deterioration of the capability which caused by the increase of the noise have been kept within limits and it have better performance than the decorrelation multiuser detector. Futher more, the advantage of the linear MMSE multiuser detector will more obvious when the noise is comparative greatly.The interference canceller multiuser detector is a kind of sub-optimal multiuser detector. The basic approach of the serial interference canceller multiuser detector is as follows: The multiuser detector arrange all users by the power of their received signals. First, It demodulate the user which have the greatest power and then minus the reconstructed interference signal from the total received signal. Second, it wipe off the secondary greatest signal and wipe off all interference signal in turn untill demodulate the last user. the multiuser detector have some advantages such as have low complexity of its structure and low calculation complications. At same time, it have some disadvantages like that there must be great influence to the next step if the estimation is not exactly this step and we must rebuild the sequence when the power of user signals change.People do a great deal of work to enhance the detection performance of the serial interference canceller multiuser detector. for example , Duel-Hallen put forward a decorrelation decision feedback multiuser detector which conjoint the decorrelation detector and the interference canceller. It can obtain better performance by make full use of the difference of the user power. For overcoming the disadvantage of the great time delay and to avoid the power compositor, people also put forward a parallel interference canceller multiuser detector.In actual communication system, the multiuser detector must be work real-time for the complexity , instability and time variability of the channel in the communication system and accordingly its difficult to obtain the user informaton exactly.The adaptive MMSE multiuser detector can carry out by making use of the LMS adaptive algorithm and the adaptive decorrelation multiuser detector can use the RLS adaptive algorithm. They both can satisfy the real-time request of multiuser detector.The adaptive decorrelation decision feedback multiuser detector combine the adaptive decorrelation detector and the interference canceller technology which can obtain better performance than the adaptive decorrelation multiuser detector。The adaptive decorrelation decision feedback multiuser detector make full use of the difference of the user power, so that the greatter the difference of the user power, the better the performance of the multiuser detector. At the same time, it make use of the decorrelation in forward and the decision feedback in the backward synthetically. On condition that the feedback data is accurate, the adaptive decorrelation decision feedback multiuser detector can remove the multiple access interferences completely.Support vector machines is a kind of study method which is put forward by Vapnik according to the theories of system studies. It can attain great classification and expansion ability under the condition of very short training sequence, is an extraordinary efficient method to solve the classify problems. Support vector machines multiuser detector, which adopt the structure risk minimum principle, not only can obtain better performance than adaptive MMSE multiuser detector ,but also only need very short training sequence.Adaptive multiuser detection algorithm can do a reaction to the time variability channel. The multiuser detectors do real-time work thoughout send out training sequence and have good performance to restrain the multiple access interferences. However, under the time variability mobile communication environment, it need to send out training sequence incessantly which is a waste of the spectrum resource. So people change to do research on the adaptive multiuser detection algorithm which need not to send out training sequence namely blind adaptive multiuser detection algorithm and blind multiuser detection algorithm for short.The blind MOE multiuser detector based on LMS algorithm which uses no more information than the single user detector, actually only uses the signature waveform and timing information of the expected user , convergence to the linear MMSE multiuser detector, therefore it is a typical blind multiuser detector.Kalman filter is a minimum square error estimator of the state vectors, the blind adaptive multiuser detector based on kalman filter not only have great perfomance in convergence and the capacity of tracking channels but also can achieve high output signal interference rate. However, it have a high calculation complications. According to the weakness of the blind adaptive multiuser detector based on kalman filter, this paper put forward a improved blind adaptive multiuser detector based on kalman filter which have better performance in convergence and the capacity of tracking channels. The improved blind adaptive multiuser detector based on kalman filter also have better performance on condition of the multi-path fading channels. Simulation proved that the improved blind adaptive multiuser detector based on kalman filter is more compliant to put into practical applications.This paper links the PARAFAC model to the DS-CDMA multiuser detection problem. The PARAFAC model which is an analysis tool rooted in psychometric and chemometrics. Exploiting this link , it derives a blind PARAFAC multiuser detector with performance close to non blind MMSE multiuser detector. The proposed multiuser detector capitalizes on code, spatial, and temporal diversity combining, thereby supporting small sample sizes, more users than sensors. Interestingly, the blind PARAFAC multiuser detector does not requre knowledge of spreading codes, the specifics of multipath, DOA calibration information, or statistical independence to recover the information bearing signals. Instead, the proposed multiuser detector relies on a fundamental result regarding the uniqueness of loe rank three-way array decomposition due to Kruskal. These and other issues are also demonstrated in pertinent simulation experiments.