Improved Rate-Energy Trade-off For SWIPT Using Chordal Distance Decomposition In Interference Alignment Networks

TL;DR

This paper introduces a chordal distance decomposition method for SWIPT in MIMO interference channels, balancing energy harvesting and data rate, and demonstrating improved trade-offs over traditional interference alignment schemes.

Contribution

It proposes a systematic, computationally efficient precoding method using chordal distance decomposition to enhance the energy-rate trade-off in SWIPT systems with interference alignment.

Findings

Higher harvested energy compared to perfect IA precoders.

Energy constraints can be met with constant rate loss.

Max-SINR and MSE based IA schemes outperform others for SWIPT.

Abstract

This paper investigates the simultaneous wireless information and power transfer (SWIPT) precoding scheme for K-user multiple-input-multiple-output (MIMO) interference channels (IC), for which interference alignment (IA) schemes provide optimal precoders to achieve full degrees-of-freedom (DoF) gain. However, harvesting RF energy simultaneously reduces the achievable DoFs. To study a trade-off between harvested energy and sum rate, the transceiver design problem is suboptimally formulated in literature via convex relaxations, which is still computationally intensive, especially for battery limited nodes running on harvested energy. In this paper, we propose a systematic method using chordal distance (CD) decomposition to obtain the balanced precoding, which improves the trade-off. Analysis shows that given the nonnegative value of CD, the achieved harvested energy for the proposed precoder is higher than that for perfect IA precoder. Moreover, energy constraints can be achieved, while maintaining a constant rate loss without losing DoFs via tuning the CD value and splitting factor. Simulation results verify the analysis and add that the IA schemes based on max-SINR or mean-squared error are better suited for SWIPT maximization than subspace or leakage minimization methods.