🔢 mathematics

A Tight Channel-Capacity Lower Bound for the Simultaneous Wireless Information and Power Transfer Integrated Receiver

This paper establishes a tight channel-capacity lower bound for simultaneous wireless information and power transfer (SWIPT) integrated receivers by deriving a closed-form channel transition matrix approximation via a 4th-order Taylor expansion of the Schottky diode's I-V curve, demonstrating that a gamma input distribution yields superior capacity estimates compared to other distributions and that higher-order modeling significantly outperforms simplified 2nd-order approaches.

Original authors: Konstantinos Ntontin, Symeon Chatzinotas

Published 2026-04-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Konstantinos Ntontin, Symeon Chatzinotas

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Charging and Talking at the Same Time

Imagine you have a smartphone. Usually, you have to plug it into a wall outlet to charge it, and you use a separate antenna to talk to cell towers.

SWIPT (Simultaneous Wireless Information and Power Transfer) is a futuristic idea where your phone does both at once. It catches radio waves from the air, turns some of that energy into electricity to charge the battery, and uses the rest of the signal to read text messages or load web pages.

The paper focuses on a specific, tricky way of building this phone: the Integrated Receiver.

The Problem: The "One-Person Band" vs. The "Two-Person Band"

Most previous research imagined a phone with two separate teams working inside it:

Team Energy: A dedicated circuit just to catch waves and turn them into battery power.
Team Data: A dedicated circuit just to decode the messages.

This is like having a two-person band. One person plays the drums (power), and the other plays the guitar (data). They don't get in each other's way, so the music is clear.

However, the authors are looking at the Integrated Receiver, which is like a one-person band. This device has only one circuit that tries to do both jobs at the same time. It catches the wave, turns it into power, and simultaneously tries to read the message from that same converted signal.

The Catch: This "one-person band" is messy. The circuit used to turn radio waves into electricity (a diode) is very non-linear. It's like trying to listen to a radio station while simultaneously trying to use the radio's speaker to power a lightbulb. The act of powering the lightbulb distorts the music.

The Challenge: How Much Data Can We Send?

In the world of information theory, we want to know the Channel Capacity: What is the absolute maximum amount of data we can send through this messy, one-person-band system without it getting garbled?

Previous researchers tried to answer this, but they made a big mistake. They used a simplified model (like a 2nd-order math approximation) to describe how the diode works.

The Analogy: Imagine trying to predict the weather by only looking at whether it's "sunny" or "rainy." You miss the clouds, the wind, the humidity, and the sudden storms.
The Reality: The diode is complex. The authors argue that you need a much more detailed map (a 4th-order Taylor expansion) to understand how the signal actually behaves.

The Solution: A New Mathematical Map

The authors did two main things to solve this puzzle:

They built a better map: Instead of the simple "sunny/rainy" model, they used a complex 4th-order math formula to describe exactly how the diode bends and twists the signal. This revealed that the system is actually capable of carrying much more data than people thought.
They found the best "shape" for the signal: To get the most data out of a channel, you have to shape your input signal perfectly.
- Some people tried using a Uniform distribution (like rolling a standard die where every number is equally likely).
- Some tried a Rayleigh distribution (a common shape in wireless signals).
- The Authors' Discovery: They found that a Gamma distribution (a specific, flexible curve) is the "Golden Ticket." When you shape your signal like a Gamma distribution, you get the tightest, most accurate estimate of how much data the system can handle.

The Results: Why This Matters

The authors ran simulations (computer experiments) with these new models and found:

The "Simple" Model was wrong: The old, simplified models (using only the 2nd-order math) underestimated the capacity. They told us the system could only carry a little bit of data, but the new, complex model shows it can carry significantly more.
The "Gamma" Shape wins: If you try to send data using the old "Uniform" or "Rayleigh" shapes, you leave performance on the table. Using the optimized Gamma shape gets you much closer to the theoretical limit.
The "Tight" Bound: They didn't just guess; they provided a mathematical "lower bound" that is extremely close to the true answer. Think of it as drawing a fence around the true answer that is so close, you can almost touch it.

The Takeaway

This paper is like upgrading from a hand-drawn sketch of a city to a high-resolution GPS map.

Old View: "This integrated receiver is messy and slow; let's just use simple math to guess how fast it is."
New View: "If we look closely at the physics (using the 4th-order math) and shape our signals correctly (using the Gamma distribution), this 'one-person band' receiver is actually a high-speed data highway."

This is a huge step forward for Ambient IoT (Internet of Things). It means we can build tiny, battery-free sensors that harvest energy from the air and send back data much faster and more reliably than we previously thought possible.

1. Problem Statement

The paper addresses the fundamental limits of Simultaneous Wireless Information and Power Transfer (SWIPT) specifically for the integrated receiver architecture.

Context: Most existing SWIPT literature focuses on "separate receiver" architectures (where power harvesting and information decoding use distinct paths), which typically model the channel as a linear Additive White Gaussian Noise (AWGN) channel.
The Challenge: Integrated receivers use a single path where the rectifier (typically a Schottky diode) performs both energy harvesting and signal detection. This introduces two major complexities:
1. Non-linearity: The diode's current-voltage ( $i_d-v_d$ ) characteristic is highly non-linear.
2. Memory Effects: Low-pass filters introduce memory into the channel.
The Gap: Previous works analyzing the information-theoretic limits of integrated receivers often rely on a simplified 2nd-order Taylor expansion of the diode characteristic. This simplification is insufficient for accurate capacity estimation, especially at higher power levels, and the true channel capacity for the more accurate 4th-order model remains an open problem.

2. Methodology

The authors propose a rigorous analytical framework to derive a tight lower bound for the channel capacity of the integrated receiver.

A. System Modeling

Signal Model: The received signal $y_{DC}$ at the output of the rectifier is modeled using a 4th-order Taylor expansion of the Schottky diode's characteristic curve:
$y_{DC} \approx k_2|\tilde{x} + \tilde{n}_{th}|^2 + k_4|\tilde{x} + \tilde{n}_{th}|^4 + n_{rec}$
Where $\tilde{x}$ is the transmitted signal, $\tilde{n}_{th}$ is thermal noise, and $n_{rec}$ is rectifier noise.
Channel Characterization: The channel is defined as a non-linear mapping from the input signal power to the DC output, subject to noise.

B. Probability Transition Matrix Approximation

To compute the mutual information $I(u; y_{DC})$ , the authors needed the probability transition matrix $p(y_{DC}|u)$ (where $u$ is the input power).

Lemma 1 (Normal Approximation): They leverage the fact that a non-central chi-squared distribution (which describes the squared magnitude of the noisy signal) converges to a Normal (Gaussian) distribution as the non-centrality parameter becomes large.
Proposition 1: By applying this lemma twice (once for the squared term and once for the resulting quartic term), they derive a closed-form Gaussian approximation for the conditional PDF $p(y_{DC}|u)$ . This avoids complex Modified Bessel functions, significantly reducing computational complexity.

C. Input Distribution Optimization

Instead of assuming a standard Gaussian input (which is optimal for linear AWGN channels but not for this non-linear channel), the authors optimize the input distribution.
They select the Gamma distribution as the input distribution $p(u)$ due to its flexibility.
The channel capacity lower bound is obtained by maximizing the mutual information over the shape parameter ( $\alpha$ ) of the Gamma distribution.

3. Key Contributions

Closed-Form Approximation: The paper provides a closed-form, tight approximation for the channel's probability transition matrix by leveraging the 4th-order Taylor expansion and normal approximations of non-central chi-squared distributions.
Tight Lower Bound: It establishes a tight channel-capacity lower bound using an optimized Gamma input distribution.
Model Validation: It demonstrates that the 4th-order model is critical, contrasting it with the widely used but inaccurate 2nd-order models found in prior literature.
Distribution Comparison: It proves that the Gamma distribution yields a significantly tighter bound compared to Rayleigh or Uniform input distributions, particularly at lower Low-Noise Amplifier (LNA) gains.

4. Numerical Results

The authors validated their analytical results using numerical simulations with realistic parameters (3 GHz carrier, 10m distance, 1W transmit power).

Tightness of the Bound: The proposed lower bound (using Gamma distribution) is extremely tight, showing a gap of only 0.1 bits compared to the approximate capacity calculated via the Blahut-Arimoto algorithm.
Comparison with Other Distributions:
- Uniform Distribution: Exhibits a gap of ~0.5 bits.
- Rayleigh Distribution: Shows a notably larger gap (looser bound).
Impact of Taylor Order:
- The 4th-order model predicts a significantly higher channel capacity than the 2nd-order model.
- The 2nd-order model (used in previous works like [7] and [8]) leads to a notable underestimation of capacity, and this underestimation gap widens as the LNA gain ( $G_{LNA}$ ) increases.

5. Significance and Conclusion

Theoretical Advancement: This work moves beyond the simplistic linear or 2nd-order non-linear models to provide a more accurate information-theoretic limit for SWIPT integrated receivers, which are crucial for low-power IoT devices where separate receivers are too energy-intensive (due to Local Oscillator requirements).
Practical Implication: The results indicate that system designers must account for the 4th-order non-linearity of diodes to accurately predict system performance. Relying on 2nd-order approximations may lead to suboptimal system designs.
Future Directions: The authors suggest future work on deriving a tight upper bound, investigating the impact of memory effects (low-pass filters) more deeply, and incorporating power transfer constraints directly into the mutual information maximization problem.

In summary, the paper provides a mathematically rigorous and computationally efficient method to estimate the capacity of SWIPT integrated receivers, proving that higher-order non-linear modeling and optimized input distributions are essential for accurate performance evaluation.