A high-speed systolic array for computing third-order cumulants

The computational complexity involved in the estimation of higher-order statistics far exceeds that of conventional second-order statistics. This paper presents a parallel-processing system that speeds up the computational process of third-order cumulants. An algorithm for estimating third-order cumulants based on matrix multiplications is presented. Then, a special structure of systolic array system for matrix multiplications is developed. The structure and complexity of the array as well as the internal structure and the delay of the processing elements are discussed. The system is suitable for VLSI implementation.