About estimation of optimal number of sensors in sensor array system based on PARAFAC decomposition

Identification of multiple sources in sensor array system can be accomplished by a powerful multilinear algebra model, PARAFAC (parallel factor) decomposition. To guarantee the accuracy of the identification of close sources (the smallest angle difference is 1°) by PARAFAC algorithm, increasing the number of sensors is a solution. Another way is to select the optimal number of sensors which is mainly influenced by three factors: signal to noise ratio (SNR), the number of sources and the range of angles. In this paper, we have fitted the function between the optimal number of sensors and one of the three factors respectively based on the results obtained by PARAFAC decomposition. Then the function between the optimal number of sensors and all three factors is concluded from the three fitting functions. In different cases, we have tested that the number of sensors calculated from this function is close to the optimal number of sensors estimated by PARAFAC decomposition.