| The signal-to-interference-plus-noise ratio (SINR) performance of two linear multistage interference cancellation schemes, i.e. linear parallel interference cancellation (LPIC) and linear successive interference cancellation (LSIC), in the uplink of a long-code code division multiple access (CDMA) system is analyzed in this thesis. We first analyze these two receivers in a synchronous environment where all users' transmitted symbols arrive at the receiver synchronously. Then, we extend results from a synchronous system to a chip synchronous one.; The SINR of a receiver's output is a function of conditional mean and variance of the receiver's decision statistic. In a symbol synchronous long-code CDMA, the conditional mean and variance of the decision variable can be expressed as functions of moments of the random matrix R for the LPIC and L for the LSIC, respectively, where R is the correlation matrix of signature sequences and L is the strict lower triangular part of R. In a chip synchronous long-code CDMA, the effective correlation matrix D(i) is derived. The decision variable's conditional mean and variance are in the same form as those in a symbol synchronous CDMA, with the correlation matrix being replaced by the effective correlation matrix.; Since the complexity of these random matrix moment calculation increases rapidly with the growth of the stage index, a graphical representation of the random matrix moment is developed to facilitate the computation. Propositions are presented to relate the moment calculation problem to several well known problems in graph theory, i.e., the coloring, the graph decomposition, the biconnected component, and the Euler tour problems. It is shown that the analytic results match very well with simulated ones. |
