Interference Alignment with Partially Coordinated Transmit Precoding

TL;DR

This paper presents a novel interference alignment algorithm with partial coordination that enhances multiplexing gains in multicell downlink systems, offering a non-iterative solution with improved throughput at practical SNRs.

Contribution

It introduces a new partial coordination model for interference alignment, extending the proper system size from K ≤ 3 to K ≤ 5, and develops a non-iterative IA algorithm for these systems.

Findings

Proper IA for K ≤ 3; proposed IA for K ≤ 5.

Non-iterative, one-shot IA algorithm improves throughput.

Backhaul data rate grows linearly with K.

Abstract

In this paper, we introduce an efficient interference alignment (IA) algorithm exploiting partially coordinated transmit precoding to improve the number of concurrent interference-free transmissions, i.e., the multiplexing gain, in multicell downlink. The proposed coordination model is such that each base-station simultaneously transmits to two users and each user is served by two base-stations. First, we show in a K-user system operating at the information theoretic upper bound of degrees of freedom (DOF), the generic IA is proper when $K \leq 3$, whereas the proposed partially coordinated IA is proper when $K \leq 5$. Then, we derive a non-iterative, i.e., one shot, IA algorithm for the proposed scheme when $K \leq 5$. We show that for a given latency, the backhaul data rate requirement of the proposed method grows linearly with K. Monte-Carlo simulation results show that the proposed one-shot algorithm offers higher system throughput than the iterative IA at practical SNR levels.