Accurate electroneurographic (ENG) signal classification is a key function in biological and neural communication systems, where peripheral nerves act as noisy, bandwidth-limited information channels interfacing with implantable bioelectronic devices. In neural decoding and stimulation (ND&S) systems, ENG signals are transmitted and subsequently classified to extract task-relevant information prior to stimulation. This processing must operate in real-time on resource-constrained implantable platforms, limiting the applicability of conventional deep neural networks in molecular and biological communication settings. To address these challenges, we propose two event-driven spiking neural network (SNN) architectures for energy-efficient ENG decoding. A Parametric Leaky Integrate-and-Fire Spiking Neural Network (PLIF-SNN) employs trainable membrane time constants to model temporal dynamics, while a Parallel Spiking Neural Network (PSNN) uses parallel spiking neurons with learnable temporal kernels and reset-free dynamics for compact temporal filtering. The proposed models were evaluated on a multichannel cuff-electrode dataset recorded from nine Long Evans rats and compared with ESCAPE-NET and MobilESCAPE-NET. For single compound action potential classification, PSNN achieved 87.21% accuracy and an 84.98% macro F1-score using only 25.5k parameters, reducing model complexity by up to 99%. On 500-sample windows, PLIF-SNN reached 90.25% accuracy with 15.4k parameters, matching MobilESCAPE-NET performance with one quarter of its parameter count. These results demonstrate that event-driven SNN-based decoding enables state-of-the-art ENG information extraction with substantially reduced computational and energy costs, supporting low-power, low-latency bioelectronic platforms for bidirectional biological and multiscale communication systems.
To enable safe and reliable intra-body communication (IBC) for medical applications, ultrasonic wideband (UsWB) technology employs low-duty-cycle pulses with time-hopping spread spectrum to effectively mitigate multipath and thermal effects. However, transient waveform distortion and multipath delay spread, which are induced by the inhomogeneity of the human body, render UsWB signals difficult to detect. This paper proposes a Squeeze-and-Excitation channel attention mechanism-based hybrid convolutional neural network and bidirectional long short-term memory (CNN-SE-BiLSTM) model for the detection of UsWB signals, in which a one-dimensional convolutional neural network (CNN) is used to capture multi-scale spatial features, the attention mechanism is employed to mitigate the impact of noise through adaptive feature recalibration, and the back-end bidirectional long short-term memory (BiLSTM) network is designed to model the global temporal dependencies of pulses in complex intra-body environments. Extensive Monte Carlo simulations are conducted to compare the bit-error rate (BER) performance of various deep learning-based detection models, including CNN, LSTM, CNN-LSTM, Transformer, and the proposed CNN-SE-BiLSTM, against traditional matched filter and energy detection receivers over intra-body fading channels. Simulation results demonstrate that the CNN-SE-BiLSTM model maintains a Flash size below 1 MB with a peak RAM usage of 280 KB and achieves the lowest BER among the evaluated schemes, offering a promising detection solution for UsWB-based IBC systems.
Molecular communications (MC) are emerging paradigms that aim to transform communication at the microand nanoscale levels by utilizing molecules as information carriers. Unlike traditional communication, molecular communication uses chemical signals, which provide distinct advantages in biological contexts. This review examines important advances in molecular communication security customized for these unique communication channels. We highlight risks and challenges in molecular systems, focusing on the confidentiality, integrity, and availability security objectives that are critical for applications such as targeted drug delivery (TDD) and in vivo networks. Furthermore, we discuss potential attacks, interdisciplinary security measures, physical/MAC layer strategies, and hybrid networks. Our contributions provide a complete overview of existing research, identify critical challenges, and potential solutions to ensure secure and reliable molecular communication.
Bioengineered neuronal systems are increasingly explored as living computational substrates, yet their end-to-end communication properties remain poorly characterized beyond global activity statistics. We model a mechanosensitive spiking neuronal network as a noisy multiple-input multiple-output (MIMO) communication system to quantify how topology and substrate mechanics shape information transfer and computation. Using a mechanosensitive Izhikevich network with time-varying stiffness, we compare random, clustered, and disintegrated topologies under matched stimulation and observation noise. Communication is quantified using transfer entropy, achievable mutual information rate from binned spike observations, and information redundancy. Computation is evaluated using a receiver-level separability metric posed as a transmitter-identification task. Across noise levels, clustered networks more consistently maintain higher achievable rates and exhibit more concentrated transmitter–receiver separability than random and disintegrated substrates. These results show that topology and mechanical state jointly regulate communication and computational separability in bioengineered neuronal networks, providing design-relevant metrics for bioengineered intelligence systems.
Biogenic volatile organic compounds (BVOCs) mediate stress signaling among plants through airborne transport and plant uptake. This letter presents a unified end-to-end model integrating: (i) time-resolved stress-induced emissions calibrated using drought–herbivory interaction data; (ii) atmospheric propagation modeled as a superposition of Gaussian puffs with stability-dependent Pasquill–Briggs dispersion; and (iii) receiver-side plant uptake as a dynamic accumulation process. Results show that near-field effects can cause concentration deviations up to 2–5× when using far-field models, highlighting the need for accurate short-range propagation. At the system level, atmospheric dilution governs signal amplitude, while uptake integration enhances persistence under sustained stress, enabling reliable detection up to ∼1.5 m under moderate wind conditions. The proposed framework provides a physically consistent and biologically interpretable basis for analyzing BVOC-mediated plant communication and predicting signaling range under realistic environmental variability.
Dysregulation of lncRNA-miRNA-mRNA (LncM2) communication networks, a key mechanism of competing endogenous RNA (ceRNA) crosstalk, plays a critical role in cancer progression. Current computational approaches are limited by their dependence on predicted interactions, lack of interpretability, and inability to effectively address transcriptomic heterogeneity. To overcome these challenges, we present MDGAM-LncM2Net, a Multi-layer Deep Graph Attention Model for reconstructing directed, context-specific LncM2 communication networks from multi-omics data. Our framework combines experimentally validated RNA interactions with transcriptomic profiles through a comprehensive analytical pipeline incorporating differential expression analysis, hypergeometric testing, Spearman correlation, and Fast Causal Inference (FCI) for directional relationship inference. MDGAM-LncM2Net leverages Graph Attention Networks (GATs) to capture topological features and Bidirectional LSTMs to model temporal expression patterns, enabling interpretable survival prediction. When applied to four cancer types (LSCC, OvCa, PrCa, and GBM), the model identified 14 pan-cancer driver lncRNAs and 11 conserved prognostic gene clusters with cancer-specific communication nodes strongly associated with patient survival. The proposed framework achieves an F1-score of 90.29% and AUC of 90.60%, representing an average improvement over the state-of-art methods. MDGAM-LncM2Net provides a robust platform for elucidating LncM2 network, with significant potential for advancing precision oncology through biomarker discovery and targeted therapy development.
Superparamagnetic iron-oxide nanoparticles (SPI-ONs) have emerged as promising information carriers in the field of molecular communication (MC). However, a critical discrepancy exists between theoretical models and experimental realities: while theoretical frameworks typically assume neutral buoyancy and neglect gravitational forces, practical implementations exhibit significant gravity-induced sedimentation. This oversight leads to substantial prediction errors and system unreliability. To address this, we developed an experimental testbed featuring a vertically oriented bifurcated channel, specifically designed to isolate and quantify the impact of gravity on signal propagation. Our experimental results empirically demonstrate that sedimentation causes significant deviations in channel impulse response (CIR), contradicting the simplifying assumptions found in existing literature. Furthermore, we elucidate the physical mechanism of sedimentation based on hydrodynamic instability to clarify the factors driving this settling behavior. This work highlights the necessity of incorporating gravitational effects into system modelling to ensure accurate performance assessment in realistic MC environments.
Detecting magnetic nanoparticles with inductive sensor coils is a proven, feasible and cost-efficient method. Recently, molecular communication in fluids, primarily vascular system models, has been investigated. The one-sensor approach is useful for determining the presence of superparamagnetic iron-oxide nanoparticles (SPIONs) in the vicinity of the sensor. However, distinguishing between a small number of particles close to the sensor and a greater number further away remains problematic. To solve this, spatial detection using multiple sensors is investigated. This opens up the possibility of utilizing spatial information in a communication context, as well as enabling imaging options that display the particle distribution over the cross-sectional area of the channel. In consequence, the concentration can be determined with greater precision. The stability of the sensors is investigated, proving the reliability of reference scales for the shift to concentration relation. Nonetheless, the presented approach is subject to several limitations and factors that must be considered in future work. To conclude the presented work, it can be said that advances could be achieved in the position and concentration determination of SPIONs, as well as a structured investigation on sensor stability.
Airborne molecular communication (AMC) systems are often subjected to persistent interfering volatile organic compounds (VOCs) originating from paints and coated surfaces in indoor environments. These background chemicals affect typical non-specific sensors such as metal oxide (MOX) sensor outputs together with the actively released signaling molecules, because of cross-sensitivity, creating drifting sensor-output base-lines, and memory-dependent effects that have not yet been systematically considered in experimental AMC studies. This study experimentally investigates how a common paint solvent (n-butyl acetate) and alcohol signaling pulses (ethanol) jointly affect an AMC system with MOX sensors. Using a static chamber and a controlled-airflow setup, we examine ethanol pulses in clean air, solvent backgrounds, superimposed signal-background conditions, and different exposure sequences. The measurements show that paint-related VOCs cause slowly varying baseline shifts that reduce ethanol pulse contrast, slow down recovery, and introduce exposure sequence dependent responses. This letter provides an experimental characterization of solvent-background interference and memory effects in MOX sensor-based AMC receivers.
Diffusive molecular communication (MC) with multiple-input multiple-output (MIMO) architectures is a promising approach for improving the throughput of nanonetworks. However, high-order concentration modulation in dense MIMO systems is highly vulnerable to inter-link interference (ILI), while fully orthogonal resource allocation requires a large number of distinguishable molecule types. To address this problem, this paper proposes a joint modulation and interference-management framework that combines linear high-order concentration modulation (LHOCM) with spatial-aware adaptive graph coloring (SA-AGC). At the link level, LHOCM increases the number of bits conveyed per symbol, and Gaussian-intersection thresholds are used to account for signal-dependent noise in molecular reception. At the network level, SA-AGC converts reliability-constrained interference relationships into a conflict graph and assigns molecule types by separating strong interfering links while allowing spatial reuse among weakly coupled links. Numerical results show that the proposed framework substantially suppresses the interference-induced error floor and improves the achievable throughput under a limited molecule-type budget, compared with full-reuse and TDMA baselines.
Quantum biology and communication engineering remain only partly integrated. One is organized around mechanisms, the other around input-output channels. This paper joins them. We model a quantum-biological communication channel (QBCC) as an open quantum system: an input is encoded into a biological quantum subsystem, the state evolves under Gorini-Kossakowski-Sudarshan-Lindblad dynamics in a structured bath, and a classical readout induces a channel law W T ( y ∣ x ) . When is such a system a channel at all, rather than an observed quantum process? An operational criterion answers this, built on an arbitrary-but-fixed encoding and a commuting abstraction-representation diagram. The channel law then connects to mutual information, capacity, quantum Fisher information, and a noise-assistance index. Four reusable primitives follow: radical-pair receivers, exciton routers, proton-tunnelling genetic error channels, and ion-coherence links. One illustrative network and four literature-informed case studies instantiate them. A penalized likelihood-ratio test is then applied to two of them. It separates a phase-sensitive interference channel, not reproduced by the sign-blind rate null, from the Fenna-Matthews-Olson complex, whose noise-assisted transport that null already reproduces. A falsifiability workflow, two worked applications, and a five-layer network stack complete the framework.
Composite DNA alphabets significantly enhance the information capacity of DNA-based storage systems but necessitate strict run-length-limited (RLL) constraints to prevent homopolymer-induced synthesis and sequencing errors. Existing constrained code constructions for these alphabets either rely solely on numerical capacity estimates or suffer from non-linear encoding complexity. Although recent general constrained-coding methods have been proposed, they require constraint-specific auxiliary computations that are impractical for RLL constraints over composite DNA alphabets due to the exponential growth in forbidden substrings. To address these challenges, this paper introduces the run-length semantics and decomposition (RS&D) framework, which fundamentally simplifies RLL constraints by defining how runs are extended or reset using two elementary operators. Based on this framework, we present two primary contributions: (i) the derivation of closed-form capacity expressions of minimal polynomial degree using recurrence-based ranking and the trace method; and (ii) a guaranteed lineartime sequence-replacement encoder applicable to any composite DNA alphabet. Furthermore, RS&D reorders composite DNA alphabets by efficiency and identifies a six-symbol alphabet that maximizes performance within the RS&D linear-time encoding constraints.
This paper presents a color-based microfluidic molecular communication (MC) testbed that experimentally demonstrates quadrature amplitude modulation (QAM) transmission using dual-dye signaling. Two distinct food dyes with different absorption characteristics are injected through independently controlled syringe pumps, and their concentrations are reconstructed from red, green, blue (RGB) measurements captured by a commercial-off-the-shelf camera. An absorbance-based estimation model is developed to obtain the dye concentrations, and the experimentally measured channel impulse responses (CIRs) reveal strong inter-symbol interference (ISI). We designed a two-dimensional molecular constellation in which each symbol is represented by a pair of dye concentrations, enabling 4-QAM and 16-QAM transmission within a microfluidic environment. Detection is performed using the minimum Euclidean distance (MED) and maximum likelihood sequence estimation (MLSE). Experimental results demonstrate the feasibility of applying QAM to MC within a microfluidic environment. The measured dye-concentration constellations exhibit systematic distortions and non-uniform spacing compared to conventional electromagnetic QAM, highlighting the unique physical characteristics of molecular transport. Using these experimentally obtained constellations, MED and MLSE detectors are evaluated, and MLSE is shown to reduce symbol error ratio (SER) by leveraging the channel memory introduced by microfluidic ISI.
Molecular communication enables bio-nanomachines to coordinate their behavior using signaling molecules and provides a foundation for collective behavior. This paper proposes a dynamic cluster formation system in which bio-nanomachines form, split, and merge micro-clusters through local interactions. Simulation studies show that micro-cluster formation improves the accuracy of sensing concentration gradients of signaling molecules and that the system exhibits three distinct phases: disordered, multi-cluster, and single-cluster. Furthermore, in environments with multiple targets, bio-nanomachines may form multiple spatially distributed micro-clusters, enabling parallel target exploration. These findings may contribute both to applications such as tumor-targeted drug delivery and to advancing our understanding of collective behavior of biological systems.
Biological intelligence provides a powerful paradigm for developing novel computational and communication frameworks inspired by living systems. Addressing complex diseases such as cancer, which remains one of the most persistent global health challenges, requires intelligent and targeted therapeutic strategies capable of distinguishing and eliminating malignant cells while preserving surrounding healthy tissue. In this work, we propose a biologically inspired navigation framework that combines chemotaxis, a natural process that governs directional cell movement in response to chemical gradients, with entropy exploration to guide nanoscale medical agents (NMA) toward cancerous regions within the tumor microenvironment (TME). The hybrid chemotaxis–entropy mechanism emulates adaptive decision making observed in biological organisms, enabling autonomous localisation and tumor treatment through targeted drug delivery. Using hypoxia as a biomarker, due to the elevated oxygen consumption of cancerous cells, numerical simulations are performed for both single- and multi-tumor configurations. Comparative analyses in multiple performance metrics reveal that the hybrid navigation approach achieves higher efficiency, reliability, and tumor elimination rates compared to random and pure chemotaxis strategies. Furthermore, the feasibility of biologically grounded navigation is demonstrated to be significantly enhanced through an entropy-guided perspective, highlighting its potential for intelligent control and communication in nanoscale systems. This study establishes a computational framework linking biological intelligence, molecular communication, and nanoscale drug delivery, laying the foundation for future bio-inspired optimisation of drug localisation in cancer therapy.
In this work, we study a three-dimensional heterogeneous mobile molecular communication (MC) system. We propose a stochastic diffusivity-based model in soft matter, where, for the time range smaller than the typical diffusivity correlation time, we consider the non-Gaussian Brownian relative displacement distribution between communicating devices. We propose the subordination approach, which analytically maps the non-Gaussian relative displacement distribution between communicating devices. We characterize the subordination approach for the stochastic diffusivity model by the channel impulse response (CIR) and derive its mean. The channel characterization of the proposed subordination approach is verified through its superstatistical approximation. For the time range beyond the diffusivity correlation time, the relative displacement between communicating devices follows a Gaussian Brownian distribution. To capture this behavior, we propose a generalized approximation model and characterize it through its CIR and mean. We consider the discrete-time statistical channel model at a high inter-symbol interference regime and derive the bit error rate expressions for the superstatistical approximation of the subordination approach and for generalized approximation. We also give the channel capacity analysis for the considered system model for both the approximation models using Lagrange multipliers. Furthermore, we show the degree of accuracy through root mean square error for the Poisson and Gaussian distributions for superstatistical and generalized approximation models.
DNA storage offers the advantages of massive capacity, long-term stability, and low power consumption relative to conventional storage media, making it a promising next-generation medium for digital information such as images. However, existing schemes face significant hurdles in coding density, precise biochemical constraint control, and biosafety assurance. To achieve more reliable and secure DNA storage image reconstruction, this work proposes an encoding framework integrating cryptographic hash primitives, state-driven dynamic mapping, and a dual-layer concatenated error-correction architecture. The framework leverages the pseudo-randomness and ”avalanche effect” of hash functions to regulate sequence generation, fundamentally eliminating base distribution periodicity. Experimental results demonstrate that the proposed method not only generates DNA sequences strictly adhering to stringent GC content (47%–55%) and homopolymer length (≤ 3) constraints but also achieves a net information density of 1.619 bits/nt. Furthermore, by actively suppressing long-range Open Reading Frames (ORFs), the scheme significantly enhances the biochemical inertness of the sequences. At a high error rate of 2%, the system maintains image reconstruction with a high SSIM value and no structural decimation. These results indicate that the proposed encoding scheme effectively guarantees the efficiency, reliability, and biosafety of the DNA storage system.
