Noncoherent Relaying in Energy Harvesting Communication Systems

Peng Liu, Saeed Gazor, Il-Min Kim, Dong In Kim

Abstract

This paper presents a unified framework for noncoherent relaying within energy-harvesting (EH) wireless networks. By leveraging Simultaneous Wireless Information and Power Transfer (SWIPT), the proposed system enables relays to scavenge energy from source signals, eliminating the need for external power sources or power-intensive channel state information (CSI) estimation. We introduce a low-complexity Gauss-Legendre Detector (GLD) that approximates optimal maximum-likelihood performance. Our findings reveal that EH-based noncoherent systems can outperform traditional grid-powered counterparts while achieving full spatial diversity.

Introduction

In modern wireless relay networks, node longevity is often constrained by limited battery capacity. Energy Harvesting (EH) from radio-frequency (RF) signals offers a sustainable solution, particularly through Simultaneous Wireless Information and Power Transfer (SWIPT). However, a significant hurdle in current SWIPT research is the reliance on coherent detection. Coherent schemes require accurate Channel State Information (CSI) at both the relay and destination. The process of estimating CSI—sending training symbols and performing complex signal processing—consumes substantial energy, often counteracting the benefits of harvesting.

This work addresses this challenge by proposing noncoherent Amplify-and-Forward (AF) relaying. By removing the requirement for CSI at any node, the system maximizes the energy available for data transmission and reduces hardware complexity, making it ideal for low-power Internet of Things (IoT) applications.

Proposed Protocols: Power Splitting and Time Switching

The authors develop a unified mathematical model for two primary noncoherent EH relaying protocols, illustrated in the figure below.

Schematic diagrams of (a) Power Splitting (PS-NcAF) and (b) Time Switching (TS-NcAF) protocols for a multi-relay network.
  1. Power Splitting (PS-NcAF): As shown in Figure (a), the relay splits the received RF signal into two streams. A fraction $\rho$ of the power is diverted to the energy harvester, while the remaining $1-\rho$ is used for information detection (ID) simultaneously.
  2. Time Switching (TS-NcAF): As shown in Figure (b), the relay alternates between harvesting and processing. A dedicated time slot of duration $\alpha T$ is used for EH, and the remaining $(1-\alpha)T$ is partitioned for information transmission across the relays.

These protocols are evaluated across two modulation schemes: $M$-ary Frequency-Shift Keying (FSK) and Differential Phase-Shift Keying (DPSK).

Technical Contributions: The Gauss-Legendre Detector

A core contribution of this research is the derivation of the Maximum-Likelihood Detector (MLD) for noncoherent EH relaying. While the MLD is mathematically optimal, it requires the evaluation of complex integrals, which is computationally prohibitive for simple EH nodes.

To bridge the gap between optimality and practicality, the authors propose the Gauss-Legendre Detector (GLD). By applying Gauss-Legendre quadrature to approximate the MLD integrals, they derived a closed-form detector. The GLD offers nearly identical Symbol Error Rate (SER) performance to the MLD but with significantly reduced computational overhead, facilitating real-time implementation on low-power hardware.

Key Insights and Performance Analysis

The system was validated using Monte Carlo simulations under Rayleigh fading conditions. Several critical insights emerged:

  • Optimal Relay Placement: Unlike traditional grid-powered systems where relays are often placed near the destination to reduce noise, EH relays must be positioned closer to the source. This "Source-Proximity Rule" ensures the relay harvests sufficient energy to power its transmission.
  • Protocol Scalability: While PS-NcAF performs better in single-relay setups, TS-NcAF is superior for multi-relay networks. In TS-NcAF, the harvested power at each relay scales with the number of relays ($K+1$), enhancing the reliability of the second-hop link as the network grows.
  • Energy Efficiency: For higher-order constellations ($M \ge 8$), $M$-FSK proves more energy-efficient than $M$-DPSK, providing a gain of approximately 4 dB at specific error thresholds.
  • Diversity and Convergence: The proposed detectors achieve full spatial diversity (order $K+1$). Furthermore, the performance gap between Amplify-and-Forward (AF) and Decode-and-Forward (DF) logic largely disappears in these noncoherent EH systems, as the primary bottleneck is the harvested energy rather than the relaying strategy.

Conclusion

This paper demonstrates that noncoherent relaying is a viable and efficient strategy for energy-harvesting networks. By eliminating CSI requirements and utilizing the low-complexity GLD, self-powered relays can match or even exceed the performance of grid-powered systems. Future research will focus on incorporating non-linear EH models and intelligent relay selection to further optimize these sustainable communication frameworks.

Summary updated

BibTeX
@article{liu2015noncoherent,
  author = {Peng Liu and Saeed Gazor and Il-Min Kim and Dong In Kim},
  title = {{Noncoherent Relaying in Energy Harvesting Communication Systems}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2015},
  doi = {10.1109/twc.2015.2462838},
}
Figure