In this paper, the author analyzed the performance of low complexity linear multi-user receiver with diversity reception, namely coherent decorrelating detector with maximal-ratio combining, and address their performance sensitivity to imperfect fading parameter estimation. To evaluate the theoretical performance of mufti-user detectors under channel mismatch conditions, the author modeled a frequency non-selective channel as the first and second order AR (Auto-Regressive process, and explore the use of Kalman filtering for channel estimation. Proposed receiver structure was shown to alleviate the near far problem due to fading and eliminate error probability floor caused by MAI (Multiple Access Interference) in a single antenna configuration, provided that perfect channel estimates are available. Results demonstrate the degradation in performance caused by imperfect channel tracking, which ultimately leads to the saturation of the error probability. Diversity proves to be an effective way of improving performance in the presence of fading and MAI. Analytical results for differentially coherent receivers are verified by the simulations.