Abstract
This paper characterizes the fundamental trade-offs between data rate, harvested energy, and reliability in wireless systems constrained by short packets and finite signal constellations. By deriving necessary conditions and proposing a novel layered code construction, the authors demonstrate the achievability of performance targets in practical Simultaneous Information and Energy Transmission (SIET) scenarios, filling a critical gap in existing communication theory.
Introduction
In the evolving landscape of the Internet of Things (IoT), low-power devices often require both data and power to be delivered wirelessly. Simultaneous Information and Energy Transmission (SIET) addresses this by using radio signals to concurrently carry messages and provide electromagnetic energy for harvesting. However, traditional SIET theory has largely relied on two unrealistic assumptions: infinite transmission time (asymptotic block-length) and infinite signal choices (Gaussian inputs).
Modern low-latency applications, such as industrial IoT, operate with short packets and discrete signal constellations (e.g., QPSK or 16-QAM). In this regime, decoding errors and energy outages are unavoidable. This work provides the first comprehensive characterization of these trade-offs under practical hardware constraints.
Positioning the Work
To understand the contribution of this research, it is essential to compare it with the existing body of literature. Most prior studies have focused on idealized scenarios that do not account for the non-linearities of real-world energy harvesters or the limitations of finite block-lengths.
As shown in the table above, this work is unique in simultaneously addressing finite channel inputs, finite block-lengths, non-linear energy harvesting (EH) models, and both Decoding Error Probability (DEP) and Energy Outage Probability (EOP).
Methodology
The authors model the system over an Additive White Gaussian Noise (AWGN) channel using a non-linear energy harvesting model. This model is more realistic than linear alternatives as it accounts for how physical diodes in harvesters respond to different signal magnitudes.
Key Technical Contributions:
- Necessary Conditions: Utilizing the concept of a "code type" (the empirical distribution of symbols), the authors derived lower bounds on DEP and EOP.
- Layered Code Construction: The researchers proposed a method to build codes by organizing constellation points into concentric "layers." This allows for precise control over energy delivery (via radii) and information rate (via the number of points per layer).
- Circular Decoding Regions: To maintain mathematical tractability while proving achievability, the authors employed disk-shaped decoding regions around constellation points.
Key Insights and Results
The research highlights a fundamental "energy-entropy conflict." To maximize information, symbols should be used with high entropy (equal frequency). Conversely, to maximize energy, high-amplitude symbols must be used more frequently. This creates a quantifiable tug-of-war between data rate and power delivery.
Major Findings:
- Information-Energy Regions: Numerical results reveal a "knee" in the performance curve; beyond a specific threshold, small increases in required energy lead to significant drops in the achievable data rate.
- Reliability Floor: In the finite block-length regime, there is a mathematically predictable level of error and outage that cannot be avoided, regardless of the coding scheme.
- Tightness of Bounds: For specific cases like BPSK, the derived lower bounds on error probability were shown to be exact, validating the theoretical framework.
Conclusion and Future Work
This work establishes the mathematical boundaries for SIET in practical, short-packet regimes. While the proposed layered code construction successfully meets target requirements, the authors note that the use of circular decoding regions introduces a small sub-optimality gap. Future research may extend these findings to fading channels, multi-antenna (MIMO) systems, and energy models that incorporate the "memory" of harvesting circuits, such as capacitor charging cycles.
Summary updated