Abstract: Due to the data congestion problem in the process of computing the eigenvalues and eigenvectors of sample covariance matrix by using centralized wireless sensor networks, a scheme which uses distributed wireless sensor networks is proposed. This scheme can compute the eigenvalues and eigenvectors of sample covariance matrix in a fully distributed way by using power method and average consensus filter related theory. Simulation results show the correctiveness and effectiveness of the algorithm.

Key words: wireless sensor network, sample covariance matrix, power method, average consensus filter