IEEE Transactions on Signal Processing
Author:
Keywords:
SISTA, Science & Technology, Technology, Engineering, Electrical & Electronic, Engineering, Array processing, Polyadic Decomposition, tensor, Vandermonde matrix, wireless communication, MIMO RADAR, MULTILINEAR-ALGEBRA, UNIQUENESS, CANDECOMP/PARAFAC, IDENTIFIABILITY, RANK, RECEIVERS, TERMS, ARRAY, Networking & Telecommunications
Abstract:
Several problems in signal processing have been formulated in terms of the Canonical Polyadic Decomposition of a higher-order tensor with one or more Vandermonde constrained factor matrices. We first propose new, relaxed uniqueness conditions. We explain that, under these conditions, the number of components may simply be estimated as the rank of a matrix. We propose an efficient algorithm for the computation of the factors that only resorts to basic linear algebra. We demonstrate the use of the results for various applications in wireless communication and array processing. © 2013 IEEE.