Comparison of eigenvector and Gram-Schmidt signal separation techniques

The use of eigenvector weighting as a means of separating signals incident on an antenna array, when a priori signal information is not available, is discussed. Analysis and modelling are used to assess and compare the performance of the eigenvector and power ratio inversion techniques. The superiority of the eigenvector technique over power ratio inversion is demonstrated for the case of two signals incident on an array with identical element patterns. The signal separation capabilities of the eigenvector and power ratio inversion techniques degrade as incident power levels approach one another. However, with eigenvector weighting there is likely to be a set of directions for which adequate signal separation is obtained, even for nearly equal incident power levels.