Two New Kinds of Interference Alignment Schemes for Cellular K-user MIMO Downlink Networks

TL;DR

This paper introduces two novel interference alignment schemes, RIR and B-DRIA, for cellular K-user MIMO downlink networks, improving degrees of freedom while addressing complexity and CSI requirements.

Contribution

It proposes two new IA schemes, RIR and B-DRIA, that reduce complexity and CSI dependence, enhancing DoF in cellular MIMO downlink systems.

Findings

B-DRIA achieves higher DoF than RIR when TAR approaches 2.

RIR outperforms retrospective IA but degrades near TAR of 1.

Both schemes outperform traditional IA in cellular MIMO scenarios.

Abstract

It is known that interference alignment (IA) plays an important role in improving the degree of freedom (DoF) of multi-input and multi-output (MIMO) systems. However, most of the traditional IA schemes suffer from the high computational complexity and require the global and instantaneous channel state information (CSI), both of which make them difficult to be extended to cellular MIMO systems. To handle these issues, two new interference alignment schemes, i.e., the retrospective interference regeneration (RIR) scheme and the beamforming based distributed retrospective interference alignment (B-DRIA) scheme, are proposed for cellular K-user MIMO downlink networks. For the RIR scheme, it adopts interference elimination algorithm to erase redundant symbols in inter-cell interference (ICI) signals, and then uses interference regeneration algorithm to avoid secondary interference. The RIR scheme obtains greater DoF gain than the retrospective interference alignment (RIA) scheme, but incurs performance degradation when the transceiver antennas ratio (TAR) approaches 1. Therefore, the B-DRIA scheme is further proposed. For the B-DRIA scheme, the cellular beamforming matrix is introduced to eliminate the ICI, and meanwhile distributed retrospective interference alignment algorithm is adopted to align inter-user interference (IUI). The simulation results show that the B-DRIA scheme obtains larger DoF than the RIR scheme locally. Specifically, when TAR approaches 1, two schemes obtain the same DoF. While TAR approaches 2, the DoF of the B-DRIA scheme is superior than the RIR scheme.