[Paper Review] Power Splitting for Full-Duplex Relay with Wireless Information and Power Transfer

Hongwu Liu, Kyeong Jin Kim+1 more|arXiv (Cornell University)|Apr 18, 2015

Energy Harvesting in Wireless Networks20 references4 citations

TL;DR

This paper proposes dynamic power splitting schemes for full-duplex relays that harvest energy from RF signals to support simultaneous information and energy transfer. By optimizing power splitting ratios using full or partial channel state information (CSI), the scheme minimizes outage probability, with the partial CSI-based approach achieving near-optimal performance while reducing feedback overhead, especially when the relay is near the destination.

ABSTRACT

This paper investigates power splitting for full-duplex relay networks with wireless information and energy transfer. By applying power splitting as a relay transceiver architecture, the full duplex information relaying can be powered by energy harvested from the source-emitted radio frequency signal. In order to minimize outage probability, power splitting ratios have been dynamically optimized according to full channel state information (CSI) and partial CSI, respectively. Under strong loop interference, the proposed full CSI-based and partial CSI-based power splitting schemes achieve the better outage performance than the fixed power splitting scheme, whereas the partial CSI-based power splitting scheme can ensure competitive outage performance without requiring CSI of the second-hop link. It is also observed that the worst outage performance is achieved when the relay is located midway between the source and destination, whereas the outage performance of partial CSI-based power splitting scheme approaches that of full CSI-based scheme when the relay is placed close to the destination.

Motivation & Objective

• To address the challenge of energy-constrained full-duplex relays in wireless networks by enabling simultaneous information and energy transfer via RF signals.
• To minimize outage probability in full-duplex relay networks degraded by strong loop interference and limited energy availability.
• To design power splitting schemes that dynamically adapt to channel conditions using either full or partial CSI, improving spectral and energy efficiency.
• To evaluate the performance trade-offs between feedback overhead and outage performance, particularly in scenarios with limited CSI knowledge.
• To determine optimal relay placement and power splitting ratios that minimize outage under varying interference and SNR conditions.

Proposed method

• Employs a power splitting (PS) architecture at the full-duplex relay, dividing the received RF signal into information and energy harvesting streams using a power splitting ratio ρ.
• Models the relay as an amplify-and-forward (AF) node that uses harvested energy to amplify and retransmit the source signal, with residual loop interference modeled as a function of the loop channel f.
• Derives the harvested energy as a function of ρ, source power, path loss, and channel gains, using energy conversion efficiency η.
• Optimizes the power splitting ratio ρ using full CSI (knowledge of h, g, f) and partial CSI (knowledge of h and f only) to minimize outage probability.
• Uses a high-SINR approximation to derive the effective signal-to-interference-plus-noise ratio (e-SINR) for performance analysis under dynamic PS.
• Employs a RTS/CTS-based channel estimation protocol to enable CSI acquisition at the relay for dynamic adaptation.

Experimental results

Research questions

• RQ1 How does dynamic power splitting based on full CSI compare to fixed power splitting in terms of outage performance under strong loop interference?
• RQ2 Can a partial CSI-based power splitting scheme achieve near-optimal outage performance while reducing feedback overhead?
• RQ3 How does relay location affect outage probability, and does the performance gap between full and partial CSI schemes vary with relay position?
• RQ4 What is the impact of loop interference and source SNR on the performance gain of dynamic over fixed power splitting?
• RQ5 Under what conditions does the partial CSI-based scheme outperform or approach the full CSI-based scheme?

Key findings

• The full CSI-based power splitting scheme achieves approximately 5.5 dB, 4.5 dB, and 3.5 dB INR gain over fixed ρ = 0.7, 0.5, and 0.3, respectively, at an outage probability of 10⁻¹.
• The partial CSI-based scheme achieves outage performance very close to the full CSI-based scheme, especially when the relay is located near the destination, with a performance gap of less than 1 dB in high-SNR scenarios.
• The worst outage performance occurs when the relay is positioned midway between the source and destination, due to balanced path loss and interference.
• At an outage probability of 10⁻², the full CSI-based scheme achieves a 4.5 dB SNR gain over fixed ρ = 0.7, 3 dB over ρ = 0.5, and 1.8 dB over ρ = 0.3.
• When the relay is near the destination, the partial CSI-based scheme slightly outperforms the full CSI-based scheme due to the dominance of loop interference and harvested energy in determining e-SINR.
• The partial CSI-based scheme incurs significantly less feedback overhead than the full CSI-based scheme, making it more practical for real-time systems with limited CSI feedback.