Performance of an Orthogonal Diversity Combining Detector for Spread-Spectrum Signals

A subspace-based method for detection of spread-spectrum signals is described in this paper that is similar to superresolution algorithms such as MUSIC. Sample vectors of despread symbol data are collected over time to form a correlation matrix, which is decomposed using eigenvalue decomposition (EVD). When using linear phase modulation, the despread symbol data contains multipath peaks that are due to a number of physical paths which dominate the channel impulse response. In some cases, however, multipath gains have some substantial amount of correlation, indicating that a method of decorrelation prior to combining would be beneficial. Bit error rate curves show that the performance of this method has some advantage over a simple RAKE receiver at low fading rates. However, EVD-based algorithms are computationally complex compared to the RAKE receiver. By implementing the EVD-based orthogonal diversity-combining detector for phase modulated signals on an Intel Pentium 4, it is shown in this paper that the update rate for this algorithm on a modern (1.1 GHz PIV) processor is sufficient for real-time operation