Interference alignment and cancellation on the MIMO interference networks

Abstract

In this dissertation, interference alignment and interference cancellation in wireless interference networks are considered. There are three major parts. In the first part of this dissertation, an interference alignment problem is presented. The optimal number of DoF K/2 for the K-user interference channel is based on the assumption that time (or frequency) extension asymptotically goes to infinity. The DoF result on the M × N MIMO interference channels is also based on that assumption. For the constant coefficient channels, only a few scenarios are analyzed by examples. From this motivation, we analyze the number of DoF for the M × N MIMO interference channels with constant channel coefficients. The DoF analysis is based on the properness condition of a interference network. Several properties of the proper systems are derived, and the DoF upperbound is computed for the K-user M × N MIMO interference channels with constant channel coefficients. It will be shown that the total DoF is bounded above as follows. The total DoF upperbound is K min(M,N) if K ≤ R where R = ⌊max (M,N)/min (M,N) ⌋, and it is max (M,N) if K = R + 1. If K ≥ R + 2, the total DoF upperbound is K(M+N)/(K+1) when M+N/K+1 is an integer, and it is ⌈K(M+N)/(K+1) ⌉ − 1 when M+N/K+1 is not an integer. In the second part of this dissertation, a new interference cancellation technique is proposed. The interference network where K synchronous co-channel users desire to transmit a base station is considered. And each user uses a space-time block code to improve system performance and to achieve higher data rate. A new zero-forcing based interference cancellation and ML decoding technique which exploits the structure of the space-time block code is proposed. The proposed technique makes it possible to apply ML decoding, and it has lower computational complexity than the previous interference cancellation scheme. Computing the orthonormal basis of the null space of target matrix can be performed by linear processing. The special characteristics of the target matrix lead to a simple form of basis vectors and a simple channel model after interference suppression. The performance of the proposed scheme is improved by ordering as in other interference cancellation schemes. Interference alignment and interference cancellation are two mainly studied interference management techniques, and the combination of two techniques may be expected to improve network performance in compared to the scenarios where neither interference alignment nor interference cancellation applies alone. From this motivation, in the third part of this dissertation, a combined interference alignment and cancellation scheme is proposed. An interference network of quasi-static 2-user X channel with a multiple-antenna relay is considered. In that scenario, the proposed interference alignment and cancellation scheme makes the received signal of each receiver an Alamoutis space-time block code. The main idea of the proposed scheme is the cancellation of aligned signal and to making the Alamoutis structure at each receiver. From that, a higher diversity gain is achieved while the DoF of 4/3 is also achieved.