Generalized Tensor Contractions for an Improved Receiver Design in MIMO-OFDM Systems
Kristina Naskovska, Martin Haardt, André Lima Férrer de Almeida
Abstract
Tensor contraction is a multilinear algebra operator that defines an inner product between two tensors with compatible dimensions. In this work, we show that the MIMO-OFDM (Multiple-Input Multiple-Output - Orthogonal Frequency Division Multiplexing) received signal can be modeled by means of the tensor contraction operator. This tensor model is obtained without requiring additional spreading and provides a new, compact, and flexible formulation of a MIMO-OFDM system. Moreover, exploiting it at the receiver side facilitates the design of several types of receivers based on iterative LS (Least Squares) or recursive LS. We compare the proposed iterative and recursive LS based receivers with and without enumeration and show their advantages over the traditional ZF -FFT (Zero Forcing - Fast Fourier Transform) receiver for MIMO OFDM. This structured tensor model also opens new research directions. Moreover, our generalized tensor contraction formulation can be extended to different multi-carrier MIMO systems.
BibTeX
@inproceedings{icassp2018_generalizedtenso,
title = {Generalized Tensor Contractions for an Improved Receiver Design in MIMO-OFDM Systems},
author = {Kristina Naskovska and Martin Haardt and André Lima Férrer de Almeida},
booktitle = {ICASSP 2018},
year = {2018}
}