Linear memoryless multiuser detection of quasi-synchronous CDMA

To date, all spread spectrum code division multiple access (CDMA) schemes have been partitioned into two categories. Either synchronous CDMA, where all users are both chip and bit synchronized, or asynchronous CDMA, where all users are neither chip nor bit synchronized relative to each other. Between these two extremes lies quasi-synchronous relative to each other. Between these two extremes lies quasi-synchronous CDMA (QS-CDMA) where users are not chip synchronized but are approximately bit synchronized. Quasi-synchronous CDMA arises in indoor radio channels where the combined propagation time and delay spread produces a round trip delay time that is limited to a few chips.;This work examines multiuser detection of QS-CDMA signals at a central base station. The primary goal is to find a low complexity multiuser detector that is near-far resistant. Complexity is an issue since QS-CDMA's primary application is in indoor systems where the quantity of base stations make cost a major consideration. Complexity is kept to a minimum by the use of linear multiuser detectors, by minimizing the amount of side information required, and by memoryless detection.;The detector is separated into three stages. The first stage filters and samples the received signal to produce a set of discrete time samples that are processed by the following stages. Stage two forms a linear combination of the samples according to a selected set of criteria to produce a soft decision estimate of the transmitted data. The criteria are zero-forcing (decorrelating), minimum mean squared error (MMSE), maximum signal to noise ration (SNR), complexity, and side information dependence. These criteria are applied separately and in combinations to produce ten detectors. Stage three makes hard decisions upon the soft decision variable to form an estimate of the transmitted bits.;The normalized least mean squares (NLMS) algorithm is used to adapt the MMSE detectors to their stationarity point. The convergence rate of the NLMS algorithm in a multiple user environment under various spreads of the users' powers is estimated and simulated. It is found that the convergence rate is a strong function of the users' power spread.;The asymptotic efficiency of the decorrelating detectors and the bit error rate of the MMSE detectors are used to quantify their performance. Their performance is evaluated for various spreading codes and the best spreading code set is identified.