Wireless-powered relaying with finite block-length codes

Mahdi Haghifam, Behrooz Makki, Masoumeh Nasiri-Kenari, Tommy Svensson, Michele ZorziarXiv 2016 · arXiv

Introduction

In traditional wireless communication theory, researchers often rely on Shannon’s capacity formula, which assumes codewords are infinitely long. While this is a helpful abstraction, it fails in the real world of the Internet of Things (IoT) and tactile internet, where packets must be short to keep latency low. Short packets (finite block-length) are more prone to errors and do not reach the theoretical Shannon limit.

Furthermore, many IoT devices are "wireless-powered," harvesting energy from radio frequency (RF) signals. This creates a complex trade-off: the relay must decide how much of the incoming signal to use for harvesting energy versus how much to use for decoding information. This paper bridges the gap by analyzing how these energy-harvesting constraints interact with the physics of short-packet communications.

System Model and Methodology

The authors focus on an Amplify-and-Forward (AF) relay system, where a relay node receives a signal, boosts it using harvested energy, and sends it to the destination. They evaluate two primary protocols:

  1. Time Switching Relaying (TSR): The relay alternates between harvesting energy and transmitting data over different time slots.
  2. Power Splitting Relaying (PSR): The relay splits the incoming signal's power, using a portion for energy harvesting and the rest for data transmission simultaneously.

To analyze performance, the study utilizes Finite Block-length (FBL) information theory. The authors derive several mathematical proofs (Lemmas) that provide tight, closed-form approximations for outage probability and throughput, validated through Monte Carlo simulations under Rayleigh fading conditions.

Key Insights

The "Sweet Spot" for Packet Length

A significant finding is that for a fixed amount of information, there is an optimal codeword length ($L$). If the packet is too short, the error rate (outage) skyrockets due to the lack of redundancy. If it is too long, the effective throughput drops because the time spent on a single packet outweighs the reliability gains. This bell-curve relationship allows engineers to optimize packet sizes for maximum efficiency.

Throughput vs. total channel uses L showing the optimal codeword length for TSR and PSR protocols.

Outage Probability and SNR Scaling

In standard relay networks with a fixed battery, the outage probability usually drops linearly as signal power (SNR) increases ($1/SNR$). However, in these wireless-powered systems, the outage drops more slowly—specifically at a rate of $\frac{\log(SNR)}{SNR}$. This occurs because the relay’s transmission power is intrinsically tied to the quality of the first-hop link. The derived analytical formulas (Lemmas 2, 3, and 4) were shown to be extremely accurate when compared to numerical simulations, particularly in medium-to-high SNR regimes.

Outage probability vs. SNR, comparing numerical results with analytical approximations.

Protocol Performance

For a wide range of settings, the Time Switching (TSR) protocol achieved a lower outage probability than Power Splitting (PSR). However, PSR can outperform TSR in specific throughput scenarios depending on the packet length and the specific power-splitting factor $\alpha$ or time-sharing factor $\theta$ chosen.

Conclusion and Future Directions

This research provides a rigorous framework for designing low-latency IoT networks. While the study focuses on AF relays and simple three-node topologies, it sets the stage for future work in Decode-and-Forward (DF) systems and multi-hop networks. Additionally, future models could incorporate hardware non-linearities in energy harvesting circuits to further refine these theoretical benchmarks.

Summary updated

BibTeX
@misc{haghifam2016wireless,
  author = {Mahdi Haghifam and Behrooz Makki and Masoumeh Nasiri-Kenari and Tommy Svensson and Michele Zorzi},
  title = {{Wireless-powered relaying with finite block-length codes}},
  year = {2016},
  eprint = {1611.05995},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1611.05995},
}
Figure