Molecular communication (MC) leverages molecules as information carriers, offering advantages such as biocompatibility and low energy consumption. Currently, MC’s research focuses on signal detection using chemical sensors, nanoparticles or biomolecules. However, challenges remain in accurately demodulating sequences of bits, particularly due to the influence of system parameters such as channel length, background flow rate, and transmitter-side actuation settings, including injection volume and valve control timing. To address these challenges, this paper introduces a MC testbed based on particle speed detection, which transforms molecular signals into particle speed signals for communication. Using hydrogen peroxide (H2O2) as the information carrier, the signal is demodulated by mixing the solution at the receiving end with specially prepared active particles and detecting the particle movement speed. Sequential transmission experiments were conducted to analyze the effects of various parameters on system performance. Experimental results demonstrate that the system accurately transmits information within a tested range, validating the theoretical model and highlighting its potential for microscopic communication applications.

Transformer models have become a cornerstone of various natural language processing (NLP) tasks. However, the substantial computational overhead during the inference remains a significant challenge, limiting their deployment in practical applications. In this study, we address this challenge by minimizing the inference overhead in transformer models using the controlling element on artificial intelligence (AI) accelerators. Our work is anchored by four key contributions. First, we conduct a comprehensive analysis of the overhead composition within the transformer inference process, identifying the primary bottlenecks. Second, we leverage the management processing element (MPE) of the Shenwei AI (SWAI) accelerator, implementing a three-tier scheduling framework that significantly reduces the number of host-device launches to approximately 1/10 000 of the original PyTorch-GPU setup. Third, we introduce a zero-copy memory management technique using segment-page fusion, which significantly reduces memory access latency and improves overall inference efficiency. Finally, we develop a fast model loading method that eliminates redundant computations during model verification and initialization, reducing the total loading time for large models from 22 128.31 ms to 1041.72 ms. Our contributions significantly enhance the optimization of transformer models, enabling more efficient and expedited inference processes on AI accelerators.

Unikernel is a promising runtime for serverless computing with its lightweight and isolated architecture. It offers a secure and efficient environment for applications. However, famous serverless frameworks like Knative have introduced heavyweight component sidecars to assist function instance deployment in a non-intrusive manner. But the sidecar not only hinders the throughput of unikernel function services but also consumes excessive memory resources. Moreover, the intricate network communication pathways among various services pose significant challenges for deploying unikernels in production serverless environments. Although shared-memory based communication on the same server can solve the communication bottleneck of unikernel-based function instances. The situation where malicious programs on the server make the shared memory untrustworthy limits the deployment of such technologies.We propose UKFaaS, a lightweight and high-performance serverless framework. UKFaaS leverages the advantages of customized operating systems through unikernel and it non-intrusively integrates sidecar functionality into the unikernel, avoiding the overhead of sidecar request forwarding. Additionally, UKFaaS innovatively implements data communication between unikernels in the same server to eliminate VM-Exit bottlenecks in RPC (remote process call) based on VMFUNC without relying on memory sharing. The preliminary experimental results indicate that UKFaaS can realize 1.8×-3.5× request throughput per second (RPS) compared with the advanced serverless system FaasFlow, UaaF and Nightcore in the Google online boutique microservice benchmark.

The NR TDD network adopts large bandwidth and high frequency band networking, and the characteristics of high frequency band will lead to large 5G wireless pathloss. On the uplink, the UE transmit power and the antenna port configuration of the terminal is less than those of the base station. In the process of services, users may encounter service difficulties due to the limited uplink coverage performance. This document describes a coverage and capacity enhancement solution. The coverage enhancement solution is based on MR and MDT data from the n28 band, which indicates the limitation scenarios of the n41 band. The MR and MDT sampling points have been collected in the maintenance platform as well as intra-system and different frequency points and RSRP of the 5G cell in the n28 band. The capacity enhancement solution is based on the overlay scenario of the NR TDD network with strong master control, the uplink time slot is added to the class A cell and the uplink capacity can be improved without interfering with class B cell.

Traditional indoor distribution systems in 2G, 3G, and 4G networks are mostly single-path, but the indoor distribution systems in 5G networks require dual paths to get greater capacity by space division multiplex. This document describes the reconstruction from a traditional single-path indoor distribution system to a dual-path indoor distribution system on a 5G network. Suitable scenarios need to be found in which the engineering workload is small and the cost is low, and in the scenario, a key point is that one path in the dual-path indoor distribution system would be leaked signals from nearby floors. If the level of the leaked signal is too low and the signal quality is poor, only low-order modulation modes can be used. Even if dual-stream access is provided, the download rate cannot be improved. The download rate is even lower than that of a single-path indoor distributed system. The suitable sicario in which leaked signal can support RANK 2 is discussed in the paper that 5G users can get higher throughput in the dual path upgrade indoor distribution system.

This paper presents a new solution for improving the dynamic response performance of Power Transmission System for Wind Turbine. A center follows control set model predictive control (CFCS-MPC) of dual active bridge converter is proposed for stabilization and rapidly response between wind turbine and output loads, which features two validation steps. First to mitigate the computational burden with a reduced prediction horizon and variable step size control set selection; then to correct the potential deviation with compensation loop and improved cost function. The operating principle of the novel CFCS-MPC is introduced for developing the cost function, which plays a vital role in voltage regulating as well as resonance damping and sampling noise resistance. Comparisons with other control methods are also provided. The theory of the proposed method has been verified in MATLAB/Simulink and shows a good application prospects in improving the dynamic response performance.