Compound Poisson Noise Sources in Diffusion-based Molecular Communication

TL;DR

This paper models biological external noise sources in diffusion-based molecular communication as a compound Poisson process, analyzing their impact on system performance and proposing optimal detection strategies.

Contribution

It introduces a novel compound Poisson noise model for biological noise sources in DMC and analyzes their effect on system detection performance.

Findings

CPNS significantly affects DMC performance predictions.

A single-threshold detector is optimal under high-rate CPNS conditions.

Conventional Poisson noise models may overestimate system performance.

Abstract

Diffusion-based molecular communication (DMC) is one of the most promising approaches for realizing nano-scale communications for healthcare applications. The DMC systems in in-vivo environments may encounter biological entities that release molecules identical to the molecules used for signaling as part of their functionality. Such entities in the environment act as external noise sources from the DMC system's perspective. In this paper, the release of molecules by biological external noise sources is particularly modeled as a compound Poisson process. The impact of compound Poisson noise sources (CPNSs) on the performance of a point-to-point DMC system is investigated. To this end, the noise from the CPNS observed at the receiver is characterized. Considering a simple on-off keying (OOK) modulation and formulating symbol-by-symbol maximum likelihood (ML) detector, the performance of DMC system in the presence of the CPNS is analyzed. For special case of CPNS in high-rate regime, the noise received from the CPNS is approximated as a Poisson process whose rate is normally distributed. In this case, it is proved that a simple single-threshold detector (STD) is an optimal ML detector. Our results reveal that in general, adopting the conventional simple homogeneous Poisson noise model may lead to overly optimistic performance predictions, if a CPNS is present.