Abstract
This paper investigates the achievable data rates for a binary two-hop communication system where the relay node relies solely on energy harvested from the source signal. By modeling the relay's finite battery as a hidden channel state and employing advanced coding techniques such as block Markov coding and timing channels, the authors derive single-letter rate expressions. The study highlights the impact of energy harvesting costs and noise distribution on the overall system capacity, providing a theoretical foundation for energy-efficient IoT and 6G architectures.
Introduction
In the push toward 6G and massive IoT, energy efficiency is paramount. One promising solution is Simultaneous Wireless Information and Power Transfer (SWIPT), where devices harvest energy from the very radio frequency (RF) signals used for communication. This paper addresses a specific, challenging architecture: a binary two-hop channel where a relay node has no external power source and must harvest energy from the source's "1" symbols to forward data to a destination.
System Architecture and the Energy Constraint
The system model consists of a transmitter ($X_1$), a relay ($Y_2:X_2$), and a receiver ($Y_3$). The relay is equipped with a finite battery and suffers from energy loss—specifically, it might take receiving multiple "1"s from the source (parameter $m$) to gather enough energy to transmit a single "1" to the destination.
This setup introduces "memory" into the channel because the relay's ability to transmit at any given moment depends on its past harvesting history (its battery state). Since the final receiver does not know the relay's internal battery level, decoding the message becomes a significant theoretical hurdle.
Modeling the Relay Battery as a State
The most powerful idea presented is treating the relay's physical energy constraint as a classic "channel with state" problem. The relay's battery level is modeled as a state $u$ in a Markov chain. Transitions between these states are governed by the received signal $y_2$ (which provides energy) and the transmitted signal $x_2$ (which consumes energy).
Proposed Achievability Schemes
The authors propose two primary schemes to find the lower bounds of the communication rate:
- State-Dependent Superposition Coding: For cases where noise exists in the second hop (or both), the authors use Block Markov Coding. Messages are sent in blocks, and the relay's transmission in one block depends on the message it decoded in the previous one. They employ backward decoding at the receiver, allowing the destination to decode the message without needing to know the exact sequence of the relay's battery states.
- Equivalent Timing Channel Approach: For scenarios where noise only exists in the first hop, the authors use a "timing" trick. Instead of looking at the bits themselves, they analyze the time intervals between transmitted "1"s. This converts a complex channel with memory into a simpler, memoryless "timing channel" where the energy harvesting process acts as additive noise.
Key Insights and Results
- The Cost of Energy ($m > 1$): Unlike simpler models, this paper assumes a realistic energy imbalance. If $m=3$, the relay must receive three "1"s to afford sending one "1". This creates a fundamental trade-off between sending information and "feeding" the relay.
- Noise Location Sensitivity: The paper reveals that noise in the first hop (Source $\to$ Relay) is significantly more damaging than noise in the second hop (Relay $\to$ Destination). This is because first-hop noise corrupts both the information and the energy harvesting process simultaneously.
- Single-Letter Expressions: Despite the complexity of channels with memory, the authors successfully derived "single-letter" expressions—mathematical formulas that are easy to compute and analyze.
Numerical simulations show that the achievable rate drops significantly as the energy cost ($m$) increases or the battery size ($U$) decreases. The authors also extended their model to include "random energy loss," where the amount of energy harvested from a "1" is a random variable, providing a more robust model for real-world hardware.
Conclusion and Future Work
This work establishes foundational achievable rates for energy-harvesting relays with finite batteries. However, finding a tight upper bound (the absolute theoretical capacity) remains an open challenge due to the difficulties of applying standard converse proofs to channels with memory. Future research may explore how a direct Source $\to$ Destination link would change the achievable rate region and extend the timing channel approach to larger battery sizes ($U > m$).
Summary updated