Comparison of Decorrelation, Minimum Mean Square Error, and Traditional Multiuser Detectors

This article provides an in-depth analysis of decorrelation and minimum mean square error (MMSE) concepts, comparing them with traditional multiuser detectors. Decorrelation is a signal processing technique widely used in communication systems to isolate target signals in complex multiuser environments. The implementation typically involves matrix inversion operations to orthogonalize user signals, such as applying a decorrelating filter computed through the inverse of the correlation matrix. MMSE serves as a fundamental performance metric for evaluating detection algorithm accuracy, where the detector minimizes the expected squared error between estimated and true symbols. Through comparative analysis of decorrelation and MMSE detectors against conventional approaches like matched filters, we examine trade-offs in computational complexity, near-far resistance, and bit error rate performance. Code implementations often leverage linear algebra libraries for matrix operations, with decorrelation requiring O(K³) complexity for K users while MMSE detectors incorporate noise variance estimation. This comparison enables better understanding of practical applicability in real-world scenarios such as CDMA systems and massive MIMO configurations.