PaperlayerView exact revision on arXiv

Source-linked AI summary

Digital Particle Communication under Flow: Frequency Analysis and Pulse Design

Fatih Merdan, Ozgur B. Akan

arXiv:2610.29567v1eess.SP

TL;DR

Particle releases for different symbols can overlap at the receiver, complicating reliable detection. The paper develops a digital communication framework with frequency analysis and pulse design that accounts for interference from both earlier and later releases. Its results relate flow speed to channel delay and bandwidth and show that optimized pulse shape and BER performance depend on noise, symbol duration, and receiver processing.

  • Problem

    Particle releases can affect observations associated with other symbols, creating intersymbol interference in particle communication.

    Source: §I · p. 1

  • Method

    The paper connects transport, pulse shaping, receiver filtering, and detection in a discrete-time framework and designs nonnegative, unit-energy pulses using a packet-level SINR criterion that includes both precursor and postcursor interference.

    Source: §I · pp. 1–2

  • Results

    BER results show that pulse shaping can substantially improve symbol-by-symbol detection, while its additional benefit depends on the equalization method.

    Source: §VI · p. 9

  • Takeaways & Limitations

    Optimized pulses spread across more of the symbol interval when noise is substantial and become more concentrated as noise decreases.

    Source: §V · p. 8

Abstract

from arXiv · show

In particle communication, particles released for different symbols can reach the receiver at overlapping times, making reliable detection difficult. This paper develops a digital communication framework that connects particle transport, pulse shaping, and receiver processing for three-dimensional free-space diffusion-advection channels. We derive the continuous-time and discrete-time frequency responses and explain how flow speed affects channel delay and bandwidth. Under advection-dominated conditions, the group delay approaches $R/V$, and the one-sided $3$-dB bandwidth scales as $V^{3/2}$. We design nonnegative, unit-energy transmit pulses using a packet-level SINR criterion that includes interference from both preceding and subsequent symbols under additive white Gaussian receiver noise. Simulations validate the analytical approximations and show that optimized pulses spread across more of the symbol interval when noise is substantial and become narrower as noise decreases. BER evaluations across SNR and symbol duration compare optimized, rectangular, and single-impulse pulses with several detection methods, identifying the operating conditions in which transmit-pulse design improves reliability.

I. Introduction

The paper frames temporal dispersion and intersymbol interference as central challenges in particle communication, then connects particle transport to digital communication tools through a unified link representation. It models diffusion-advection transport and receiver noise to support frequency analysis, pulse shaping, and detection.

I. Introduction

  • Particle releases can affect observations in other symbol intervals, creating intersymbol interference and making channel memory relevant to reliable communication.The work accounts for both precursor and postcursor interference relative to a delayed decision reference.Source: §I · pp. 1–2
  • The framework connects physical transport, transmit-pulse shaping, receiver filtering, and symbol detection in a unified discrete-time representation of the particle communication link.It is developed for three-dimensional free-space diffusion-advection channels.Source: §I · p. 1
  • The pulse design uses a packet-level SINR criterion for a common nonnegative, unit-energy transmit pulse, accounting for receiver noise and both precursor and postcursor interference.The paper compares optimized pulses with rectangular and single-impulse pulses using detection and equalization methods.Source: §I · p. 2

II. Channel Discretization

  • The framework assumes sufficient sampling and computation, and models dominant receiver electronic noise as additive white Gaussian noise.The noise approximation is motivated by thermal noise in the receiver circuitry.Source: §II · p. 2
  • The transport model represents particle concentration using diffusion, advection, and a source term, with the flow field and diffusion coefficient capturing medium and application differences.The analysis uses a common mathematical framework rather than specializing to a particular scale or application.Source: §II · p. 2

III. Frequency Domain Analysis of the Constant Flow

The constant-flow analysis derives the channel's frequency response and relates its delay and bandwidth to flow speed. Under advection-dominated conditions, delay is approximately the transit time and bandwidth scales superlinearly with flow speed.

III. Frequency Domain Analysis of the Constant Flow

  • Under advection-dominated conditions, group delay is approximately constant and equal to the advective transit time R/V.The approximation is close for stronger flow but departs from the exact, frequency-dependent delay as flow speed decreases.Source: §III · pp. 3–4
  • For fixed R and D, the effective bandwidth scales approximately as V^3/2, indicating a superlinear increase with flow speed under advection-dominated conditions.Source: §III · p. 4
  • The approximate discrete-time frequency response neglects shifted spectral copies when they do not overlap significantly over the principal frequency interval.This condition permits neglecting aliasing terms with nonzero shifts.Source: §III · p. 4
  • The continuous-time and approximate discrete-time responses show that stronger flow increases channel bandwidth and yields a more nearly linear phase response.Weaker flow produces more pronounced phase nonlinearity in the plotted configuration.Source: §III · p. 4

A. End-to-End Signal Representation

The end-to-end representation connects transmit pulses and the diffusion-advection channel to receiver filtering and symbol-rate detection. It expresses the sampled link as a symbol-rate model with noise.

A. End-to-End Signal Representation

  • The figures compare normalized channel impulse responses and continuous-time and approximate discrete-time frequency responses across constant flow speeds.Both captions specify v2 = v3 = 0.Source: Figure 1 · p. 4, Figure 2 · p. 4
  • The transmit pulse is convolved with the discrete-time channel, and receiver filtering produces the effective pulse used in the sampled link.The effective pulse after the channel is g[n] = h[n] ∗ p[n], while x[n] includes receiver filtering.Source: A. End-to-End Signal Representation · pp. 4–5
  • The sampled symbol-rate link is represented as y = Xs + v, with X capturing the effective pulse-channel response and v the sampled receiver noise.The model follows receiver filtering and downsampling at the selected symbol timing.Source: A. End-to-End Signal Representation · p. 5
  • The receiver-filtered noise is generally colored, whereas the sampled symbol-rate noise is white Gaussian.The symbol-rate noise autocorrelation is σ²δ[q].Source: A. End-to-End Signal Representation · p. 5

B. Pulse Design for Particle Communication

The pulse-design formulation chooses a nonnegative, unit-energy transmit pulse and a time-reversed receiver filter. Its packet-level objective accounts for desired-signal power, ISI across symbol lags, and receiver noise.

B. Pulse Design for Particle Communication

  • The design constrains the transmit pulse to be nonnegative and unit energy, and uses its time reversal as the receiver filter.This choice is motivated by matched-filter detection for a known deterministic signal in AWGN.Source: B. Pulse Design for Particle… · pp. 5–6
  • The packet-level objective counts each lagged ISI coefficient Nsym − |ℓ| times, reflecting how often it contributes within a finite symbol packet.The desired contribution is x0s_i; contributions at other lags are ISI.Source: B. Pulse Design for Particle… · pp. 5–6
  • The pulse is designed by minimizing an inverse-SINR objective under the nonnegativity and unit-energy constraints.The objective includes packet-level desired contribution, ISI power, and receiver noise.Source: B. Pulse Design for Particle… · p. 6

C. Detection Methods

The paper considers four established digital-communication detection methods for the particle-communication framework: symbol-by-symbol detection, maximum-likelihood sequence detection, LMMSE equalization, and DFE.

C. Detection Methods

1) Symbol-by-Symbol Detection under an ISI-Free Model:

The symbol-by-symbol baseline keeps only the desired-symbol term in its model and makes independent decisions, without compensating for the ISI present in the actual received signal.

1) Symbol-by-Symbol Detection under an ISI-Free Model:

2) Maximum Likelihood Sequence Detection:

The detector models observations with a finite-window ISI matrix and selects the symbol sequence that best matches the observations under the Gaussian noise model. Viterbi computes this ML solution when retained memory is short enough, while pilot requirements and state growth limit its practical range.

2) Maximum Likelihood Sequence Detection:

  • The ML sequence detector selects the candidate sequence minimizing the squared residual between observations and the finite-window ISI model.The model retains precursor and postcursor ranges, and the observation noise is Gaussian with variance σ_w^2.Source: 2) Maximum Likelihood Sequence Detection · p. 6
  • Viterbi computes the same finite-window ML solution when retained memory is sufficiently short, avoiding exhaustive testing of |A|^N_sym sequences.With precursor ISI, an observation is scored only after the required future symbols have been hypothesized.Source: 2) Maximum Likelihood Sequence Detection · p. 6
  • The trellis has |A|^(L−+L+) states and requires at least M pilot symbols to initialize from known symbols preceding the data.The retained memory is M = L− + L+; exact Viterbi MLSE is practical only for sufficiently short retained memory.Source: 2) Maximum Likelihood Sequence Detection · p. 6
  • For long-memory channels, MLSE is a theoretical benchmark, while LMMSE equalization and DFE are more computationally tractable alternatives.Source: 2) Maximum Likelihood Sequence Detection · p. 6

3) Linear MMSE Equalization:

The section presents mean-aware LMMSE estimation for nonzero-mean symbol amplitudes and a DFE that uses centered symbols and prior decisions to address postcursor terms. The feedforward coefficients are selected by an MSE criterion, with feedback coefficients chosen to cancel considered postcursor contributions.

3) Linear MMSE Equalization:

  • LMMSE estimation accounts for the nonzero mean of symbol amplitudes by using an affine estimator with centered observations and symbols.For i.i.d. symbols and noise, the estimate simplifies to a regularized linear system with λ = σ_w^2/σ_s^2 before mapping estimates to the nearest constellation points.Source: 3) Linear MMSE Equalization · pp. 6–7

4) Decision Feedback Equalization:

  • The DFE centers each symbol and observation around the transmitted-symbol mean, then maps the estimate plus that mean to a symbol decision and recenters it for feedback.Source: 4) Decision Feedback Equalization · p. 7
  • Assuming correct past decisions, DFE feedback coefficients are chosen from the feedforward coefficients to cancel the considered postcursor terms.The feedback coefficients satisfy c_r = −q_r for the retained postcursor lags.Source: 4) Decision Feedback Equalization · p. 7
  • The DFE feedforward taps are obtained by minimizing the estimation-error MSE, yielding a linear system for the taps.Source: 4) Decision Feedback Equalization · p. 7

V. Simulation Results

The simulations validate the channel approximations and show that pulse and receiver choices affect BER differently across channel memory and symbol duration. Optimized pulses broaden with higher noise, while stronger flow supports low BER at shorter symbol durations.

V. Simulation Results

  • The group-delay approximation agrees best at the strongest tested flow, while the bandwidth approximation approaches the exact curves as flow rises but fails at very low speeds.At very low speeds, exact bandwidth approaches a diffusion-controlled limit; the approximation continues to decrease.Source: §V · p. 7, Figure 4 · p. 7
  • Pulse design uses numerical constrained optimization with multiple feasible initializations, and the BER experiments assume perfect timing knowledge.The sampling offset is fixed near the advective transit delay; timing acquisition and synchronization-error robustness are left for future research.Source: §V · p. 7
  • At a nominal 10 dB, high-memory channel coefficients span several neighboring symbols on both sides of the desired symbol, whereas low-memory coefficients cluster closer to it.The paper labels the V = 0.005 m/s and 0.03 m/s channels high-memory and low-memory, respectively, relative to the common symbol duration.Source: §V · pp. 7–8, Figure 5 · p. 8
  • Optimized pulses spread across more of the symbol interval at low SNR, then narrow as SNR rises, with high-memory pulses remaining broader than low-memory pulses.At high SNR, the low-memory pulse approaches an impulse-like shape; broader pulses are favored when noise is substantial.Source: §V · p. 8, Figure 6 · p. 8
  • In the high-memory channel, symbol-by-symbol detection stays near BER 0.5; in the low-memory channel, the optimized pulse reaches approximately 10^-2 at 0 dB versus about 0.4 for the rectangular pulse.Block LMMSE gives the largest improvement in the high-memory channel; with ISI-aware detection, rectangular and optimized pulses perform closely over much of the SNR sweep.Source: §V · pp. 8–9, Figure 7 · p. 8, Figure 8 · p. 9
  • Longer symbol durations reduce neighboring-response overlap and improve BER; stronger flow reaches low BER at shorter durations, while the single-sample impulse remains less effective at 0 dB.Rectangular and optimized pulses benefit substantially from equalization as symbol duration increases.Source: §V · p. 9, Figure 9 · p. 9

VI. Conclusion

The framework links transport, frequency characterization, pulse design, and detection for three-dimensional free-space diffusion-advection channels. It relates flow and noise conditions to channel behavior and reports when pulse shaping improves detection.

VI. Conclusion

  • Under advection-dominated conditions, group delay approaches R/V and one-sided 3-dB bandwidth scales as V^3/2.These channel properties connect flow speed to delay and bandwidth.Source: §VI · p. 9
  • Nonnegative, unit-energy pulse design accounts for precursor and postcursor ISI, with optimized pulses broader at substantial noise and more concentrated as noise decreases.The pulse shape balances desired-signal gain against ISI reduction.Source: §VI · p. 9
  • BER results show pulse shaping can substantially improve symbol-by-symbol detection, but its added benefit depends on equalization method and channel memory.The reported BER-versus-symbol-duration figure covers v1 = 0.1 m/s at SNR = 0 dB over 200 Monte Carlo trials.Source: §VI · p. 9, Figure 10 · p. 10
  • Increasing SNR alone was insufficient to overcome severe ISI, whereas stronger flow enabled reliable communication at shorter symbol durations.Source: §VI · p. 9
Loading 2610.29567v1…