| Multiuser detection and space-time processing, which can effectively suppress multiple access interference (MAI) and inter-symbol interference (ISI), and alleviate "near-far" effect problem inherent in CDMA systems, have been regarded as two key techniques in the third-generation (3G) and future 4G mobile systems. The two advanced signal processing techniques have the potential to enable high capacities suited for Internet and multimedia services, and also dramatically increase range and reliability. Aiming at complexity simplification, robustness, blind adaptive implementation, and parallel implementation of neural networks, this dissertation is focused on multiuser detection and adaptive space-time receiver combining multiuser detection with smart antenna. The main contributions and original ideas included in this dissertation are summarized as follows.1. The performance of bit error rate, asymptotic multiuser efficiency, near-far resistance, local near-far resistance, complexity, and processing delay of decorrelating multiuser detector alleviating edge effect (DMUDAEE) is quantitatively analyzed. Furthermore, we analyze the expression of average partial correlation coefficients and give its numerical computation method in asynchronous DS-CDMA systems. Relative delay of the maximal and minimal correlation coefficients is also solved by evolution algorithm. Based on the obtained results, a simulation method for evaluating the performance of multiuser detection on the condition of average, maximal, and minimal partial correlation coefficients is proposed. The performance of the DMUDAEE is evaluated by the method. It is shown that theoretic analysis accords with that of Monte Carlo simulations, and the DMUDAEE with lower complexity and smaller processing delay takes on outstanding performance in eliminating MAI and "near-far" resistance.2. A MMSE multiuser detector alleviating edge effect (MMSEAEE) with low complexity is proposed and its asymptotic efficiency, "near-far" resistance and computational complexity are qualitatively analyzed. The detector can relax the edge effect because of truncation processing in asynchronous systems and obtain excellent trade-off between resisting MAI and background noise. With performance analysis and numerical simulation, it is shown that the performance of MMSEAEE approaches that of DMUDAEE at low signal-to-interference ratio (SIR) and exceeds that of DMUDAEE at high SIR.3. Based on the analysis of signature mismatch by using rectangular window and root-raised cosine pulse shaping filter on non-frequency-selective fading multipath channels, a batched robust blind linear multiuser detection is proposed. Givenuncertainty set of signature waveform, the parameter controlling robustness of the detector can be precisely calculated. Furthermore, for tracking dynamic wireless channels and reducing its complexity, we proposed an blind adaptive robust multiuser detection with the same computational complexity as the classical recursive least square algorithm by using Taylor series expansion to approximately solve the parameter controlling robustness. Performance analysis and plenty of numerical simulation shows that the robust multiuser detector and its adaptive version have excellent performance of output signal-to-interference-plus-noise ratio, bit error rate, asymptotic multiuser efficiency and robustness even existing stronger MAI. On the other hand, the blind robust adaptive multiuser detector can quickly converge.4. A blind adaptive decorrelating RAKE receiver on frequency-selective fading multipath channels is proposed, which first makes use of blind adaptive multiuser detector for a RAKE finger to resist MAI and ISI, and adaptively combines all fingers with maximum ratio criteria. The receiver does not require training sequences and estimate channel parameters in advance, and so it can be adopt in uplink and downlink. With numerical simulations and performance comparison, the proposed receiver shows outstanding performance in eliminatin...
|
PDF Full Text Request