Tensor Decomposition Based Beamspace ESPRIT

for Millimeter Wave MIMO Channel Estimation

Abstract—We propose a search-free beamspace tensor-ESPRIT

更多

algorithm for millimeter wave MIMO channel estimation. It is a

multidimensional generalization of beamspace-ESPRIT method

by exploiting the multiple invariance structure of the measurements.

Geometry-based channel model is considered to contain

the channel sparsity feature. In our framework, an alternating

least squares problem is solved for low rank tensor decomposition

and the multidimensional parameters are automatically associated.

The performance of the proposed algorithm is evaluated

by considering different transformation schemes.

Index Terms—Tensor decomposition, beamspace ESPRIT, millimeter

wave, MIMO, channel estimation

收起