Abstract
In optical wireless scattering communication, received signals are captured by a photomultiplier tube (PMT) and sampled through short but finite intervals. This process forms a signal vector used for symbol detection. This paper investigates the upper and lower bounds on the transmission rate of such systems, demonstrating that the gap between these bounds vanishes as thermal and shot noise variances approach zero. To address the computational complexity of Maximum A Posteriori (MAP) detection, a low-complexity receiver is derived using piecewise polynomial approximations. Additionally, threshold-based signal detection is analyzed using rules based on detection error probability and Kullback-Leibler (KL) distance. Numerical results show that KL distance-based thresholds are robust to selection variations at low noise levels, and the proposed piecewise cubic approximations achieve near-optimal performance.
Introduction
Non-Line-of-Sight (NLOS) optical scattering communication is a critical technology for environments where radio frequency (RF) signals are restricted or impractical, such as in hospitals, aircraft, or secure facilities. Unlike traditional line-of-sight links, scattering communication relies on the interaction of photons with atmospheric particles.
While theoretical models often assume an "ideal" photon-counting receiver capable of detecting every electron arrival in infinitesimal time, practical hardware relies on Photomultiplier Tubes (PMTs) and finite-interval sampling. This practical setup introduces two significant challenges: complex noise profiles (comprising signal-dependent shot noise and signal-independent thermal noise) and high computational requirements for optimal signal detection. This paper bridges the gap between theoretical ideals and practical implementation by modeling signal detection under short-interval sampling.
System Modeling and Transmission Rate
The researchers developed a mathematical framework for a Waveform Processing (WP) receiver. In this model, the PMT output is characterized as a combination of Poisson-distributed photoelectron arrivals and additive Gaussian noise. By analyzing the mutual information of this system, the authors derived upper and lower bounds for the achievable transmission rate.
A key theoretical insight from this analysis is the "Vanishing Gap." The study proves that the gap between the upper and lower transmission rate bounds approaches zero as thermal and shot noise variances decrease. This convergence confirms that the model provides a highly accurate "speed limit" for practical optical scattering links using high-quality PMT hardware. Furthermore, the results indicate that increasing the number of sampling intervals ($M$) per symbol significantly improves the transmission rate, effectively approaching the capacity of an ideal Poisson channel.
Signal Detection Strategies
Piecewise Polynomial Approximations
Optimal detection typically employs the Maximum A Posteriori (MAP) criterion. However, the Log-Likelihood Ratio (LLR) calculations required for MAP detection involve complex logarithmic and exponential operations that are computationally prohibitive for high-speed digital signal processors.
To mitigate this, the authors proposed piecewise linear and piecewise cubic polynomial approximations. These functions are designed to mimic the optimal MAP detector's decision surface while remaining computationally efficient.
- Piecewise Linear: Offers the lowest complexity but shows performance degradation in high-noise environments.
- Piecewise Cubic: Identified as the "sweet spot" for practical design. It provides a negligible performance loss compared to the exact MAP detector, even when thermal noise is significant.
Threshold-Based Detection and KL Distance
For simpler receiver architectures, the paper explores threshold-based detection. This involves selecting a hard threshold to determine the presence of a signal. The researchers proposed two selection rules based on detection error probability and the Kullback-Leibler (KL) distance.
Interestingly, the KL distance analysis revealed that the optimal detection threshold is remarkably robust. For small noise variances, the KL distance curve remains relatively flat near its peak. This suggests that engineers can select a threshold from a wide range of values without a significant drop in performance, making the system more resilient to hardware fluctuations and calibration errors.
Conclusion
This research provides a robust framework for understanding and optimizing practical optical wireless scattering communication. By proving the tightness of transmission rate bounds and offering low-complexity piecewise cubic approximations, the authors provide a path toward high-speed, real-time receivers that do not require the extreme computational overhead of traditional MAP detection.
Future work may extend these findings beyond On-Off Keying (OOK) to higher-order modulation schemes like Pulse Position Modulation (PPM). Additionally, addressing Inter-Symbol Interference (ISI) caused by "pulse stretching" in high-speed scattering channels remains a vital area for subsequent experimental validation on FPGA-based testbeds.
Summary updated