Run-Length-Limited ISI-Mitigation (RLIM) Coding for Molecular Communication

Melih Sahin, Özgür B. AkanarXiv 2024 · DOI · arXiv

TL;DR

This paper introduces RLIM, a novel coding scheme for molecular communication (MC) that significantly reduces inter-symbol interference (ISI) by minimizing the number of "1-bits" in code words. This leads to improved reliability and lower bit error rates compared to existing methods, making MC more practical for applications like in-body sensing and drug delivery.

What Problem Does This Paper Address?

Molecular communication (MC) holds promise for scenarios where traditional wireless communication is impractical, such as within the human body or in confined microfluidic environments. However, a major hurdle in MC is inter-symbol interference (ISI). This occurs because molecules released for one signal interval can linger and be detected in subsequent intervals, blurring the signal and making it difficult for the receiver to decode the intended message accurately. Existing coding schemes, while helpful, often struggle to effectively mitigate ISI, especially under the strict power and time constraints inherent in MC. This paper tackles the challenge of designing robust coding and detection mechanisms that can overcome ISI and enable reliable communication in MC channels.

Methods

The core contribution is the Run-Length-Limited ISI-Mitigation (RLIM) coding scheme. RLIM codes are designed with specific constraints:

  • Run-Length Limitation: Each "1-bit" is followed by at least i "0-bits" (where i is a parameter of the code). This directly limits how long a "1-bit" signal can persist and interfere with subsequent signals.
  • Leading Zeros: Codes start with i "0-bits".
  • At Least One "1-bit": Every code word must contain at least one "1-bit". This is crucial for certain detection schemes.

A key innovation is the minimization of the total number of "1-bits" within the codebook for a given length. This is achieved by selecting a fixed-size subset of admissible (i,∞)-RLL codes that has the fewest "1-bits". This strategy allows for a larger molecule budget per transmitted bit under standard normalizations, directly improving reliability.

For detection, the authors propose a provably optimal linear-time greedy decoder. This decoder is equivalent to Viterbi decoding with a "last-wins" tie-breaking rule but offers lower computational complexity. They also explore both adaptive and static threshold detection techniques, with the static threshold being analytically estimated.

Datasets

The paper does not introduce new datasets. Instead, it relies on extensive simulations to evaluate the performance of RLIM codes. These simulations utilize:

  • Binomial distribution models: To simulate the number of molecules detected in a signal interval.
  • Particle-tracking simulations: To model the physical movement of molecules in a more realistic, albeit computationally intensive, manner.
  • Time-varying drift channel models: To assess performance under dynamic channel conditions.

The simulations are conducted across a broad range of parameters, including different coding orders (i), block lengths (k), signal intervals (ts), molecule counts (M), diffusion coefficients (D), receiver radii (rR), distances between transmitter and receiver (ro), and receiver noise variances (σn²).

Key Insights and Ideas

  • Minimizing "1-bits" is key for MC reliability: The insight that reducing the number of "1-bits" in code words directly translates to a higher molecule budget per bit and thus lower BER is a central theme. This is a practical optimization for the physical constraints of MC.
  • A simpler, faster decoder: The proposed greedy decoder, equivalent to Viterbi with a specific tie-breaker, offers significant computational advantages without sacrificing performance. This is crucial for real-time applications.
  • RLIM outperforms classical RLL: The RLIM codes, by incorporating the "1-bit" minimization principle, consistently achieve better BER than traditional Run-Length-Limited (RLL) codes, even when those RLL codes are corrected by the same algorithm.
  • Adaptive detection is vital for dynamic channels: For channels with random or time-varying drift, adaptive threshold detection is shown to be significantly superior to static thresholds.
  • Trade-offs between code length and storage: Increasing the coding block length (k) improves BER but exponentially increases storage requirements, highlighting a practical design trade-off.

Key Results

  • RLIM codes achieve a significant BER advantage over classical RLL codes and other prominent ISI-mitigating codes across a broad range of simulation scenarios.
  • The proposed greedy decoder (equivalent to Viterbi last-wins) consistently outperforms Viterbi with first-wins or random tie-breaking in terms of BER.
  • In elementwise comparisons across various parameters, RLIM codes win a substantial majority of comparisons against RLL codes, with a weighted mean improvement of approximately 1.522x at winning points.
  • For a non-drift channel, the optimal static threshold detection method consistently outperforms the optimal dynamic threshold method.
  • In time-varying drift channels, dynamic threshold detection consistently achieves lower bit error rates.

Limitations and Future Work

  • Analytical Threshold Estimation: While the paper provides an analytical method for estimating static thresholds, it acknowledges that more accurate theoretical distributions for N<sub>average</sub> and <sup>i</sup>N<sub>average</sub> are needed for further improvements. This could potentially eliminate the need for pilot signals for threshold estimation.
  • Passive Receivers: For passive (non-absorbing) receivers, stronger ISI can narrow the gains of coding schemes, and at very short signal intervals, uncoded transmission might even yield lower BER (though overall reliability remains poor).
  • Long-Memory Drifts: The benefits of RLIM's dynamic threshold might diminish under long-memory or heavy-tailed drifts, suggesting a need for further study in these regimes.
  • Storage vs. Performance Trade-off: The exponential increase in storage requirements with increasing coding block length (k) is a practical limitation that needs careful consideration for real-world applications.
  • Future Work: The authors suggest focusing on developing more precise theoretical distributions for threshold estimation and exploring the performance of RLIM codes in more complex channel environments, such as those with blood-flow drift.

Summary updated

BibTeX
@misc{sahin2024run,
  author = {Melih Sahin and Özgür B. Akan},
  title = {{Run-Length-Limited ISI-Mitigation (RLIM) Coding for Molecular Communication}},
  year = {2024},
  eprint = {2411.15955},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2411.15955},
}
Figure