Research On Linear Unified Model And Nolinear Improved Algorithm Of Multiuser Detection For CDMA Systems

The capacity and performance of a code division multiple access (CDMA) system is essentially dominated by multiple access interference (MAI). For a long time, the suppression and cancellation of MAI is a key technology on performance improvements and capacity augment, thereinto the multiuser detection (MUD) has caught attention as a core technology.This paper presents the unified model of the linear MUD and deals with its performance and application. The nonlinear interference cancellation algorithm is analyzed and some new methods are presented. The main research works and results are listed as follows.Based on the diagonal loading method, a linear multiuser detector with parameters is presented, which is the unified model of the linear MUD. The performance of the linear unified model is discussed and the transformation between the Matched Filter (MF), the decorrelator, and the minimum mean square error (MMSE) detector is studied through the linear unified model. Based on the linear unified model, an approach to implement the MMSE receivers in multipath fading channels is presented and a linear MUD with the dual-mode is proposed. Analysis and simulation results show that the new method is equivalent to the conventional MMSE detection method but easier to implement than the latter on fading channel. The dual-mode detector can implement the detection of the decorrelator and the matched-filter simultaneously and inhibit the defect caused by the independent detection.The amplitude mismatch is inevitable in the Successive Interference Canceller (SIC) while the performance of SIC depends on the estimate accuracy of the signal reconstruction. It is very difficult to analyze and compute the performance loss caused by the amplitude mismatch directly from the Bit Error Ratio (BER) formula The approximate BER formula of SIC with amplitude mismatch is proposed by Gaussian approximate analysis method and the formula of the quantifiable analysis on the percentage amplitude mismatch, Signal-to-Noise Ratio (SNR) and performance loss is derived by the Lagrange mean value theorem, which describes quantificationally the effect of the amplitude mismatch on the performance of systems.A grouping SIC with diagonal loading detection is proposed through the interval grouping method, each stage of which can implement MMSE detection of the grouped users so that the demodulation delays is reduced and the performance of systems is improved. The spreading sequence is general used directly as the dispreading sequence for detection, however it is not completely orthogonal, which results in MAI. Based on Gram-Schmidt algorithm, this paper proposes an orthogonal dispreading SIC (OSIC) through constructing each stage dispreading sequence from the spreading sequence. Analysis and simulation show that OSIC can inhibit MAI completely and the enhancement of noise in each stage and has better BER performance than SIC. A multipath successive interference cancellation algorithm based on RAKE technique is proposed for the multipath fading scenario. A nested RAKE receiver is presented by this new algorithm. Theoretical analysis and simulation results show that the new RAKE receiver with simple structure and easy implementations can suppress the multipath interference and improves the BER performance.Although the Parallel Interference Canceller (PIC) is simple in structure and has good ability to mitigate the near/far effect, its performance is affected by the estimate accuracy of the initial data and signal power. The interference estimate performance between the Hard PIC (HPIC) and the Linear PIC (LPIC) is analyzed and the performance comparison between LPIC and MF is studied, and the application condition of SNR is proposed quantificationally for LPIC and MF.In order to improve the interference estimate of LPIC in multipath fading channels, a Weight-cancelled PIC (WPIC) is presented, the weighted factor for different stages of which is obtained by MMSE criterion. Theoretical analysis and simulation results show that WPLIC has robustness to near-far effect and outperforms LPIC significantly.The effect of the decision sort on the decision-feedback multiuser detector with feed forward transform is analyzed theoretical and a new method of decision sort is presented based on the iteration estimate of the signal amplitude A Switch Decision-Feedback multiuser Detector (SDFD) is proposed, which can select the detection model between the matched filter and the decorrelator according to the channel change through a decision switch strategy. Theoretical analysis and simulation results show that SDFD can be suitable for low signal noise ratio scenario and has better BER performance than the decorrelaing decision-feedback multiuser detector and the decision-feedback multiuser detector with the orthogonal feed forward.