Vessel Network Topology in Molecular Communication: Insights from Experiments and Theory

For the arXiv Version of the Paper: click here Short Description of the Paper: In this work, we present a novel analytical and physical channel model for molecular communication (MC) in vessel networks, validated with measurements in a fluidic testbed featuring branched channel topologies. The model accounts for key transport mechanisms, including advection, molecular diffusion, eddy diffusion, and reversible adsorption at the channel walls. It is general in nature and supports arbitrary signaling molecules and transparent receivers (RXs). To enable experimental validation, we specialize the model to an MC system which uses superparamagnetic iron-oxide nanoparticles (SPIONs) as signaling molecules. By deriving an analytical model for a planar coil inductive RX and integrating it with the channel model, we obtain a complete end-to-end model for SPION-based MC. The testbed used for validation features the injection of SPIONs via a venous cannula and a micropump, with continuous background flow maintained by a gear pump. SPIONs propagate through the channel topology made of silicon tubing and are detected at the planar coil inductive RX due to their magnetic susceptibility. Experimental validation includes both straight channels, consisting of a single pipe between the injection site and the RX, and branched channels with two alternative paths of varying length. For each configuration, repeated channel impulse responses (CIRs) are recorded and compared to model predictions. Building on the theoretical framework, we introduce topology-dispersion metrics that directly link the topological channel structure of a vessel network to the signal-to-noise ratio (SNR) at the RX. The relevance and accuracy of these metrics are further supported through experimental validation. Data Description: All measurements used in the paper are provided in the file experimental_data.zip. The data includes 97 CIRs, 54 from straight channel topologies and 43 from branched channel topologies. The data is organized by channel topology as follows: Top-Level Structure:The dataset is divided into two main folders: straight_channelsContains measurements for five different straight channel lengths. branched_channelsContains measurements for four distinct branched channel topologies. Subfolder Structure:Within each of the two main folders, there are subfolders for the distinct channel topologies. Each of these subfolders in turn includes the following .csv files and folders: resonance_frequency_raw.csvRaw sensor output over time, showing the resonance frequency in Hertz during repeated SPION injections. pump_activation_raw.csvRaw micropump activation signal over time, indicating the timing of each SPION injection. resonance_frequency_shift_realizationsFolder containing the individually segmented CIRs. Typically, 11 CIRs are included per topology. Each signal represents the resonance frequency shift over time, relative to the sensor's baseline frequency. measured_average_resonance_frequency_shift.csvThe ensemble-averaged CIR computed from the individual realizations in the folder above. All .csv files are comma-separated and have the following format: Column 1: Time (in seconds) Column 2: Measured values (either in Hertz or Volts, depending on the file) Contact: If you have any questions or suggestions for improvements, feel free to contact us: Timo Jakumeit ([e-mail]) Lukas Brand ([e-mail]) Sebastian Lotter ([e-mail])