Dynamic Range Compression Using Hadamard Processing and Decorrelation Spreading

We present an efficient usage of quantization bits applied to parallel input signals that are correlated each possessing average power varying over a large dynamic range. Our solution utilizes Hadamard-matrix processing (HMP) which consists of decorrelation spreading followed by the Hadamard matrix operation. If the input signals are uncorrelated, the Hadamard-matrix operation transforms the parallel signal sources to have the equivalent average power prior to quantization, thereby effectively minimizing the overall number of quantization bits required to achieve system performance. To insure that the signals are uncorrelated, we implement decorrelation spreading prior to the Hadamard. Different decorrelator methods of varying complexity are examined. We show that the decorrelation methods when used with the Hadamard matrix can influence the immunization of "signal clipping" due to quantization in addition to realizing significant system capacity and/or power gains. We demonstrate the performance of the HMP on the mobile-user objective system (MUOS) through simulation by comparing overall G/T degradation for different scenarios