Molecular communication (MC) is a bio-inspired paradigm for information exchange that leverages the properties of messenger molecules for data transmission. Recognized as a promising physical-layer technique for the Internet of Bio-Nano Things within biological entities, MC facilitates intricate collaboration and networking among micro-scale devices. Data-driven detectors are favored in MC receivers due to their complex and dynamic channel characteristics. The messages of the MC systems are vital, while the recovery process via the data-driven detectors mostly exhibits an opaque nature. To address this transparency issue, this letter uses an artificial intelligence tools, SHapley Additive exPlanations (SHAP), to explain the basic principles of data-driven detectors in MC from a systematic perspective. Through this approach, important feature points of the received signals are extracted, which further enhances the detection performance of MC with a reduced need for signal samples, thereby substantiating the role of interpretability in improving the functional capabilities of MC systems.

This paper establishes a molecule harvesting transmitter (TX) model in molecular communication (MC). In particular, we consider that molecules are encapsulated in vesicles generated within the TX and released from the TX through membrane fusion process. We also consider that the TX membrane is covered by heterogeneous receptors of varying sizes and at arbitrary locations, where the receptors can absorb the released molecules once the molecules hitting any of the receptor. Assuming that the vesicle generation follows a jump process, with each vesicle generated at distinct time instants, and assuming a transparent receiver (RX), we calculate the molecule release rate, the expected fraction of absorbed molecules at the TX, and the received signal at the RX. All obtained analytical expressions are functions of all receptors' locations and sizes, and are validated by particle-based simulations. Our numerical results indicate that evenly distributed receptors on the TX membrane absorb more molecules than randomly distributed receptors or a single receptor. Furthermore, inspired by the biological phenomenon that cells can regulate their release of new molecules by interacting with the molecules that are already present in the environment, we incorporate a negative feedback mechanism (NFM) at the TX. This mechanism utilizes the number of molecules absorbed by the TX as a criterion to determine if the TX should stop releasing additional molecules. We then derive the closed-form expression for the expected fraction of recyclable molecules for a single emission. Here, the pool of recyclable molecules comprises both the molecules that remain unreleased by the TX due to NFM and those that are absorbed back by the TX. Our numerical results demonstrate that incorporating NFM can reduce inter-symbol interference (ISI) while maintaining the same peak received signal as without NFM. Additionally, our results show that TXs incorporating both molecule harvesting and NFM can achieve a higher energy efficiency and lower error probability than TXs employing only molecule harvesting or neither functionality.

This paper investigates over-the-air computation (AirComp) over multiple-access time-varying channels, where devices with high mobility transmit their sensing data to a fusion center (FC) for averaging. To combat the Doppler shift induced by time-varying channels, each device adopts orthogonal time frequency space (OTFS) modulation. Our objective is minimizing the mean squared error (MSE) for the target function estimation. Due to the multipath time-varying channels, the OTFS-based AirComp not only suffers from noise but also interference. Specifically, we propose three schemes, namely S1, S2, and S3, for the target function estimation. S1 directly estimates the target function under the impacts of noise and interference. S2 mitigates the interference by introducing a zero padding-assisted OTFS. In S3, we propose an iterative algorithm to estimate the function in a matrix form. In the numerical results, we evaluate the performance of S1, S2, and S3 from the perspectives of MSE and computational complexity, and compare them with benchmarks. Specifically, compared to benchmarks, S3 outperforms them with a significantly lower MSE but incurs a higher computational complexity. In contrast, S2 demonstrates a reduction in both MSE and computational complexity. Lastly, S1 shows superior error performance at small SNR and reduced computational complexity.

This paper investigates over-the-air computation (AirComp) in the context of multiple-access time-varying multipath channels. We focus on a scenario where devices with high mobility transmit their sensing data to a fusion center (FC) for averaging. To combat the time-varying channel and Doppler effect, each device adopts orthogonal time frequency space (OTFS) modulation. After signals are received by the FC, the aggregated data undergoes demodulation and estimation within the delay-Doppler domain. We leverage the mean squared error (MSE) as a metric for the computational error of OTFS-based AirComp. We then derive the optimal transmit power at each device and signal scaling factor at FC for minimizing MSE. Notably, the performance of OTFS-based AirComp is not only affected by the noise but also by the inter-symbol interference and inter-link interference arising from the multipath channel. To counteract the interference-induced computational errors, we incorporate zero-padding (ZP)-assisted OTFS into AirComp and propose algorithms for interference cancellation. Numerical results underscore the enhanced performance of ZP-assisted OTFS-based AirComp over naive OTFS-based AirComp.

This paper explores the integration of device-to-device communications into clustered federated learning (FL), where clients are grouped into multiple clusters based on the similarity of their learning tasks. To mitigate communication costs, we propose an efficient FL algorithm. Specifically, we designate a primary client within each cluster responsible for uploading the model to the server, while other clients within the cluster serve as secondary clients. Each secondary client assesses its model's similarity to the primary client by computing a layer-wise model distance. If a secondary client's model distance exceeds a predefined threshold, indicating a divergence from the primary client's model, it transmits its model distance to the edge server. The primary client then updates the cluster model parameters and broadcasts them to the secondary clients within the cluster. Closed-form expressions for the time spent by the proposed layer-wise efficient FL is derived. Numerical results validate the training accuracy of the layer-wise efficient FL and demonstrate a notable reduction in communication costs compared to naive FL.