Reduce Data Transmission Research Articles

Machine learning‐based data fusion method for wireless sensor networks

AbstractFor wireless sensor networks (WSNs), sensor nodes lose a certain amount of energy during the information collection and transmission process, and sensor nodes powered by non‐replaceable batteries have limited energy and need to be controlled for energy consumption. In the face of the energy consumption issue in WSN data transmission, research has been conducted to analyze data fusion methods in order to reduce energy consumption. Based on machine learning techniques, a Deep Stacked Auto‐Encoder (DSAE) model is constructed and trained using a layer‐wise greedy approach. By combining this model with WSN, an algorithm based on the DSAE model, called Deep Stacked Auto‐Encoder Data Fusion Algorithm (DSAEDFA), is obtained to do data fusion. The results show that compared to other algorithms, the proposed fusion algorithm has better fusion performance. When the number of iterations is set to 500, the DSAEDFA has 281 surviving nodes, which is 10 more than the Back‐Propagation Data Fusion Algorithm (BPDFA) and 144 more than the Low Energy Adaptive Clustering Hierarchy (LEACH) algorithm. When the number of failed nodes is 40, the DSAEDFA has a network survival time of 2562 rounds, which is 746 rounds longer than the LEACH algorithm. The research method effectively extends the lifespan of wireless sensor networks and reduces data transmission energy consumption. Compared to previous methods, the proposed method consider the factors of node residual energy and distance on the basis of traditional routing protocols, making the selection of cluster heads more reasonable. The proposed method can organically combine the DSAE model with the clustering model, optimize the data fusion method, and improve the performance of the algorithm. In addition, by combining the DSAE model, a machine learning technique with clustering models has been expanded in terms of the application scope.

Event‐triggered H∞ and reduced‐order asynchronous filtering for fuzzy Markov jump systems with time‐varying delays

AbstractThe challenge of and asynchronous reduced‐order design for Takagi‐Sugeno (T‐S) fuzzy Markovian jump systems (MJSs) with time‐varying delays under the event‐triggered scheme (ETS) is investigated in this dissertation. A distributed event‐triggered strategy is provided on the basis of the specified triggering function to ensure consensus in the system, so effectively reducing data transmission. The existence conditions for a class of Markovian jump discrete‐time systems are determined. Unlike previous results, we present a novel membership function‐dependent fuzzy Lyapunov‐Krasovsikii (L‐K) functional with mode‐dependent integral terms, resulting in a stochastically stable filtering error system with the desired performance. By solving linear matrix inequality (LMIs), the recommended filter parameters are achieved. The proposed reduced‐order filter is demonstrated in numerical examples as effective as it is advantageous.

S-MBR: An Efficient Scaling Scheme for PM-MBR-Coded Distributed Storage System

<p>With the rapid increase in data storage, distributed storage systems need to add new nodes to reduce storage and computation pressure. However, the existing scaling schemes are not efficient enough. To solve these issues, A scaling scheme called S-MBR is proposed, which has two notable features. Firstly, S-MBR achieves uniform data distribution by using a symmetric data layout to evenly place data blocks on the expanded data nodes. Secondly, S-MBR minimizes data movement during data redistribution and parity update processes by re-locating data blocks to symmetric positions and performing partial parity block calculations after the storage nodes are expanded. S-MBR can reach the lower bound of data migration volume when the number of nodes required for reconstruction is equal to the number of nodes required for repair. According to mathematical analysis and experimental results, compared to many typical scaling schemes, S-MBR can reduce data transmission volume by up to approximately 93% and average response time by around 41%.</p> <p> </p>

Research on image processing of electric power system terminals based on reinforcement learning and mobile edge computing optimization

AbstractThis research is dedicated to the optimization of power system terminal image processing based on RL and MEC. With the continuous development of power system, the demand for image processing of terminal equipment is increasing day by day. However, traditional image processing methods have the problems of high computing complexity and real‐time and energy consumption. To solve this problem, this study introduces the idea of RL and MEC to improve the efficiency and performance of image processing of power system terminals. By modeling and optimizing the image processing task of the power system terminal equipment, the intelligent adjustment of the processing parameters is realized to adapt to the needs of different scenarios. MEC technology is introduced to move image processing tasks from the central server to the edge device, reducing data transmission delay and network burden, thus improving real‐time performance and reducing energy consumption. The experimental results show that the proposed optimization method based on RL and MEC has a significant performance improvement compared with the traditional method in the power system terminal image processing. The framework our proposed has achieved significant improvement in task completion latency, achieving higher system energy efficiency compared to traditional methods.

Intelligent data-driven condition monitoring of power electronics systems using smart edge–cloud framework

The ongoing revolution in industrial production- Industry 4.0, is driven by transformative technologies such as the Industrial Internet of Things (IIoT), Artificial Intelligence (AI), single-board computers, and 5G communication. As the trend towards IIoT continues, an increasing number of industrial drive systems and their fleets are being connected to the cloud. This enables the manufacturers to perform condition monitoring (CM) and streamlined maintenance activities. At the heart of these drive systems are Power Electronics Systems (PESs), which operate at high switching frequencies (10 kHz–1 MHz) to efficiently transfer electrical power and deliver it to a load in a controlled manner. However, due to their functionalities and the presence of semiconductor switches, PESs are susceptible to failure, necessitating effective condition monitoring (CM) for fault detection and improved lifetime. Link to this issue, to enable CM based on high-frequency data, an industrial site with multiple electric drives is required to record data up to 15TB/week. Therefore, there is a demand from industrial partners to establish intelligent communication between a fleet of physical systems and the cloud to reduce transmission, storage, and bandwidth costs, as well as to enable real-time fault detection and learning from fleet operations. This paper proposes an intelligent edge–cloud computing methodology to address the challenge of high-frequency data monitoring for PESs, focusing on novelty detection and selective data transmission to reduce transmission costs. The methodology involves developing a novel edge–cloud framework that incorporates a neural network-based novelty detector for selective data transmission from physical systems to the cloud. The proposed methodology is evaluated through hardware tests, demonstrating a significant reduction in data transmission (94%) and potential cost savings of up to €5.9k/year for a single remote system. 95.6% detection accuracy of the PQ phase is obtained during experimental tests over 590 samples. Thus, this paper contributes to the vision of the smart grid and IIoT by analyzing the Power Quality (PQ) monitoring problem of a three-phase grid and showcasing the capability of the proposed framework in terms of novelty detection and data transmission cost reduction. To conclude, the proposed intelligent edge–cloud computing methodology offers a promising solution for effective condition monitoring of PESs, with potential cost savings and improved fault detection capabilities. By leveraging advanced technologies and intelligent data-driven approaches, this framework advances the goals of Industry 4.0 and paves the way for efficient and reliable industrial operations in the digital age.

Facial emotion recognition with a reduced feature set for video game and metaverse avatars

AbstractThis paper presents a novel real‐time facial feature extraction algorithm, producing a small feature set, suitable for implementing emotion recognition with online game and metaverse avatars. The algorithm aims to reduce data transmission and storage requirements, hurdles in the adoption of emotion recognition in these mediums. The early results presented show a facial emotion recognition accuracy of up to 92% on one benchmark dataset, with an overall accuracy of 77.2% across a wide range of datasets, demonstrating the early promise of the research.

Event-triggered interval type-2 fuzzy control of networked systems with extended dissipative under Markovian switching

In this technical note, the problem of the extended dissipative control for interval type-2 fuzzy model with nonlinear network systems in the event-triggered framework is investigated. A novel dynamic event-triggered strategy is developed to ascertain the instants of activation. It has the advantage of reducing data transmission and optimising the utilisation of communication resources. On the basis of the proposed mechanism, the stochastic Lyapunov function including internal dynamic variables is constructed. Based on this function, coupled with the inequality scaling technique and the application of Dynkin's formula, we derive a sufficient condition for the system's stochastic asymptotic stability and extended dissipativity. The design of the dynamic event-triggered extended dissipative controller becomes challenging because the new sufficient condition is nonlinear for some matrix variables. However, utilising the Projection Lemma and the co-design approach, such a controller can be developed based on the obtained analysis result. The methodology is demonstrated through a rigorous numerical simulation case study.

Automatic learning-based data optimization method for autonomous driving

One of the major reasons for the explosion of autonomous driving in recent years is the great development of computer vision. As the number of model parameters increases, many large autonomous driving models need to be deployed in the cloud. The increasing data on the vehicle side has put huge pressure on transmission bandwidth. In order to reduce data transmission consumption and improve the quality of uploaded data, we design a data optimization method based on reinforcement learning. Through this method, the collected data will be filtered and only those data that are valuable for model training will be uploaded, thereby saving communication resources. The results show that this method can reduce the collected data to 42.9% by implementing data optimization, while increasing the convergence speed by 46% in model training.

An Efficient FPGA-based Depthwise Separable Convolutional Neural Network Accelerator with Hardware Pruning

Convolutional neural networks (CNNs) have been widely deployed in computer vision tasks. However, the computation and resource intensive characteristics of CNN bring obstacles to its application on embedded systems. This article proposes an efficient inference accelerator on Field Programmable Gate Array (FPGA) for CNNs with depthwise separable convolutions. To improve the accelerator efficiency, we make four contributions: (1) an efficient convolution engine with multiple strategies for exploiting parallelism and a configurable adder tree are designed to support three types of convolution operations; (2) a dedicated architecture combined with input buffers is designed for the bottleneck network structure to reduce data transmission time; (3) a hardware padding scheme to eliminate invalid padding operations is proposed; and (4) a hardware-assisted pruning method is developed to support online tradeoff between model accuracy and power consumption. Experimental results show that for MobileNetV2 the accelerator achieves 10× and 6× energy efficiency improvement over the CPU and GPU implementation, and 302.3 frames per second and 181.8 GOPS performance that is the best among several existing single-engine accelerators on FPGAs. The proposed hardware-assisted pruning method can effectively reduce 59.7% power consumption at the accuracy loss within 5%.

Combining Machine Learning and Edge Computing: Opportunities, Challenges, Platforms, Frameworks, and Use Cases

In recent years, we have been observing the rapid growth and adoption of IoT-based systems, enhancing multiple areas of our lives. Concurrently, the utilization of machine learning techniques has surged, often for similar use cases as those seen in IoT systems. In this survey, we aim to focus on the combination of machine learning and the edge computing paradigm. The presented research commences with the topic of edge computing, its benefits, such as reduced data transmission, improved scalability, and reduced latency, as well as the challenges associated with this computing paradigm, like energy consumption, constrained devices, security, and device fleet management. It then presents the motivations behind the combination of machine learning and edge computing, such as the availability of more powerful edge devices, improving data privacy, reducing latency, or lowering reliance on centralized services. Then, it describes several edge computing platforms, with a focus on their capability to enable edge intelligence workflows. It also reviews the currently available edge intelligence frameworks and libraries, such as TensorFlow Lite or PyTorch Mobile. Afterward, the paper focuses on the existing use cases for edge intelligence in areas like industrial applications, healthcare applications, smart cities, environmental monitoring, or autonomous vehicles.

An Artificial Intelligence Mechanism for the Prediction of Signal Strength in Drones to IoT Devices in Smart Cities

Abstract: Smart cities are comprised of intelligent objects that can collectively and automatically improve living standards, preserve lives, and function as a sustainable environment. Drones or Unmanned aerial vehicles (UAV), robotics, cognitive computing, and the Internet of Things (IoT) are mandatory to enhance the intellectual ability of smart cities by enhancing connectivity, energy efficiency, and Quality of Signal (QoS). Consequently, the integration of drones with IoT plays a crucial part in enabling a vast array of smart-city applications. Drones are undeniably the technology of the future. They glide in the air, keeping an eye on things in their metropolis. They do not need human control or operation. They capture data based on visuals and sounds by employing a variety of sensors, webcams, and mics, and then transfer it to a gateway for processing and retrieving information. Drones will let us gather vast volumes of data for processing, while also enhancing the intelligence of smart cities. The drone's signal is vulnerable to absorption, refraction, diffraction, and attenuation. Therefore, it is essential to predict the signal from the drone. This study presents an intelligent method using Deep Learning to predict the signal strength for enhancing network connectivity and delivering the desired QoS of IoTs and drone integration. This enables effective data transmission, boosts QoS for end-users/devices, and reduces data transmission power consumption.

Enhancing Wireless Sensor Network Lifetime through Energy-Efficient Data Clustering and Compressed Forwarding in video Processing

Wireless Sensor Networks (WSN) play a crucial role in diverse data gathering applications, but face a significant challenge in the form of limited energy reserves within sensor nodes. Enhancing the network's Quality of Service, particularly its lifetime, is paramount. Prolonging the network's operational span hinges on mitigating energy consumption, with communication accounting for a substantial portion of nodal power usage. By reducing data transmission, not only can energy consumption be curtailed, but also bandwidth requirements and network congestion can be minimized. In the context of Wireless Sensor Networks, the Distributed Similarity-based Clustering and Compressed Forwarding (DSCCF) approach strives to construct data-similar iso-clusters with minimal communication overhead. This technique involves extracting trend and magnitude components from lengthy data series using an LMS filter, resulting in what is termed data projection. Data similarity between nodes is assessed by measuring the Euclidean distance between these data projections, thereby facilitating efficient and low-overhead iso-cluster formation. To further economize intra-cluster communication, an adaptive-nLMS-based dual prediction framework is employed. During each data collection round, the cluster head holds instantaneous data for each cluster member, using either prediction or direct data communication. Furthermore, inter-cluster data is reduced via a multi-level lossless compressive forwarding technique. Impressively, this proposed approach has achieved an 80% reduction in data while maintaining optimal data accuracy for the collected information. The transmission of inter-cluster data exclusively occurs through a network backbone comprised solely of cluster heads. Initially, the cluster heads establish this network backbone. Each cluster head dispatches a link request query towards the sink through the backbone, receiving a link reply message containing path length and the weakest link of the path. The cluster head repeats this process for each available path, subsequently selecting the most optimal path based on the acquired information and its reliability in terms of link quality

Enhancing Information Maximization With Distance-Aware Contrastive Learning for Source-Free Cross-Domain Few-Shot Learning.

Existing Cross-Domain Few-Shot Learning (CDFSL) methods require access to source domain data to train a model in the pre-training phase. However, due to increasing concerns about data privacy and the desire to reduce data transmission and training costs, it is necessary to develop a CDFSL solution without accessing source data. For this reason, this paper explores a Source-Free CDFSL (SF-CDFSL) problem, in which CDFSL is addressed through the use of existing pretrained models instead of training a model with source data, avoiding accessing source data. However, due to the lack of source data, we face two key challenges: effectively tackling CDFSL with limited labeled target samples, and the impossibility of addressing domain disparities by aligning source and target domain distributions. This paper proposes an Enhanced Information Maximization with Distance-Aware Contrastive Learning (IM-DCL) method to address these challenges. Firstly, we introduce the transductive mechanism for learning the query set. Secondly, information maximization (IM) is explored to map target samples into both individual certainty and global diversity predictions, helping the source model better fit the target data distribution. However, IM fails to learn the decision boundary of the target task. This motivates us to introduce a novel approach called Distance-Aware Contrastive Learning (DCL), in which we consider the entire feature set as both positive and negative sets, akin to Schrödinger's concept of a dual state. Instead of a rigid separation between positive and negative sets, we employ a weighted distance calculation among features to establish a soft classification of the positive and negative sets for the entire feature set. We explore three types of negative weights to enhance the performance of CDFSL. Furthermore, we address issues related to IM by incorporating contrastive constraints between object features and their corresponding positive and negative sets. Evaluations of the 4 datasets in the BSCD-FSL benchmark indicate that the proposed IM-DCL, without accessing the source domain, demonstrates superiority over existing methods, especially in the distant domain task. Additionally, the ablation study and performance analysis confirmed the ability of IM-DCL to handle SF-CDFSL. The code will be made public at https://github.com/xuhuali-mxj/IM-DCL.

Ellipsoidal Trajectory Optimization for Minimizing Latency and Data Transmission Energy in UAV-Assisted MEC Using Deep Reinforcement Learning

Due to their flexible deployment and movement capability, unmanned aerial vehicles (UAVs) are being utilized as flying mobile edge computing (MEC) platforms, offering real-time computational resources and low-latency data processing for a wide range of applications. This article aims to explore a UAV-assisted MEC system where multiple UAVs provide MEC services to mobile devices (MDs) using an ellipsoidal trajectory. Depending on the position, size, and orientation of the ellipsoidal trajectories, the coverage area of the UAV, the energy consumption, and the task transmission latency of MDs change. This has rarely been investigated in the existing works. Furthermore, unlike other studies, we consider that each MD has varying task offloading rates, which, together with varying user densities, makes the problem more challenging. Therefore, we formulate an optimization problem that finds the center position, major radius, minor radius, and rotation angle of the ellipsoidal trajectory of UAV-assisted MEC servers, to minimize the total transmission latency and energy consumption of mobile devices while taking into account the required data transmission rate, task transmission time, and energy consumption constraints. Then, we transform this optimization problem into a Markov decision process and propose a deep Q-learning-based ellipsoidal trajectory optimization (DETO) algorithm, to resolve it. The results from our simulations demonstrate that DETO efficiently computes the optimal position and trajectory for each UAV, and can achieve better performance compared to other baselines, leading to the reduced data transmission latency and energy consumption of mobile devices across a range of simulation scenarios.

BioEdge: Accelerating Object Detection in Bioimages with Edge-Based Distributed Inference

Convolutional neural networks (CNNs) have enabled effective object detection tasks in bioimages. Unfortunately, implementing such an object detection model can be computationally intensive, especially on resource-limited hardware in a laboratory or hospital setting. This study aims to develop a framework called BioEdge that can accelerate object detection using Scaled-YOLOv4 and YOLOv7 by leveraging edge computing for bioimage analysis. BioEdge employs a distributed inference technique with Scaled-YOLOv4 and YOLOv7 to harness the computational resources of both a local computer and an edge server, enabling rapid detection of COVID-19 abnormalities in chest radiographs. By implementing distributed inference techniques, BioEdge addresses privacy concerns that can arise when transmitting biomedical data to an edge server. Additionally, it incorporates a computationally lightweight autoencoder at the split point to reduce data transmission overhead. For evaluation, this study utilizes the COVID-19 dataset provided by the Society for Imaging Informatics in Medicine (SIIM). BioEdge is shown to improve the inference latency of Scaled-YOLOv4 and YOLOv7 by up to 6.28 times with negligible accuracy loss compared to local computer execution in our evaluation setting.

Refactoring BZIP2 on the new‐generation sunway supercomputer

AbstractHigh‐performance computing is progressively assuming a fundamental role in advancing scientific research and engineering domains. However, the ever‐expanding scales of scientific simulations pose challenges for efficient data I/O and storage. The data compression technology has garnered significant attention as a solution to reduce data transmission and storage costs while enhancing performance. In particular, the BZIP2 lossless compression algorithm has been widely used due to its exceptional compression ratio, moderate compression speed, high reliability, and open‐source nature. This paper focuses on the design and realization of a parallelized BZIP2 algorithm tailored for deployment on the New‐Generation Sunway supercomputing platform. By leveraging the unique cache patterns of the New‐Generation Sunway processor, we propose the highly tuned multi‐threading and multi‐node implementations of the BZIP2 applications for different scenarios. Moreover, we also propose the efficient BZIP2 libraries based on the management processing element and computing processing element which support the commonly used high‐level (de)compression interfaces. The test results indicate that the our multi‐threading implementation achieves maximum speedup of 23.09 (8.57) in decompression(compression) compared to the sequential implementation. Furthermore, the multi‐node implementation achieves 50.81% (26.35%) parallel efficiency and peak performance of 16.6 GB/s (52.8 GB/s) for compression(decompression) when scaling up to 2048 processes.

Collaborative analysis for drug discovery by federated learning on non-IID data.

With the increasing availability of large-scale QSAR (Quantitative Structure-Activity Relationship) datasets, collaborative analysis has become a promising approach for drug discovery. Traditional centralized analysis which typically concentrates data on a central server for training faces challenges such as data privacy and security. Distributed analysis such as federated learning offers a solution by enabling collaborative model training without sharing raw data. However, it may fail when the training data in the local devices are non-independent and identically distributed (non-IID). In this paper, we propose a novel framework for collaborative drug discovery using federated learning on non-IID datasets. We address the difficulty of training on non-IID data by globally sharing a small subset of data among all institutions. Our framework allows multiple institutions to jointly train a robust predictive model while preserving the privacy of their individual data. We leverage the federated learning paradigm to distribute the model training process across local devices, eliminating the need for data exchange. The experimental results on 15 benchmark datasets demonstrate that the proposed method achieves competitive predictive accuracy to centralized analysis while respecting data privacy. Moreover, our framework offers benefits such as reduced data transmission and enhanced scalability, making it suitable for large-scale collaborative drug discovery efforts.

A Multi-Agent Prediction Method for Data Sampling and Transmission Reduction in Internet of Things Sensor Networks.

Sensor networks can provide valuable real-time data for various IoT applications. However, the amount of sensed and transmitted data should be kept at a low level due to the limitations imposed by network bandwidth, data storage, processing capabilities, and finite energy resources. In this paper, a new method is introduced that uses the predicted intervals of possible sensor readings to efficiently suppress unnecessary transmissions and decrease the amount of data samples collected by a sensor node. In the proposed method, the intervals of possible sensor readings are determined with a multi-agent system, where each agent independently explores a historical dataset and evaluates the similarity between past and current sensor readings to make predictions. Based on the predicted intervals, it is determined whether the real sensed data can be useful for a given IoT application and when the next data sample should be transmitted. The prediction algorithm is executed by the IoT gateway or in the cloud. The presented method is applicable to IoT sensor networks that utilize low-end devices with limited processing power, memory, and energy resources. During the experiments, the advantages of the introduced method were demonstrated by considering the criteria of prediction interval width, coverage probability, and transmission reduction. The experimental results confirm that the introduced method improves the accuracy of prediction intervals and achieves a higher rate of transmission reduction compared with state-of-the-art prediction methods.

A secure and effective data aggregation in WSN for improved security and data privacy

The increasing prevalence of internet usage and mobile devices has underscored the critical importance of safeguarding personal data. This is especially important in Wireless Sensor Networks (WSNs), where information typically requires in-network computing and collaborative processing. These computationally demanding approaches are not suitable for resource-constrained WSN nodes. Aggregating data effectively while protecting user data is a major challenge in wireless sensor networks. Many privacies of homomorphism encryption-based WSN data aggregation methods have been created and investigated recently. Since cluster leaders (aggregators) may rapidly combine cypher texts without decryption, communication overhead is reduced, making these data aggregation methods more secure than traditional ones. However, the base station only receives aggregated output, causing issues. Initial limits apply to aggregating functions. If the aggregated output is the sum of sensing data, the base station cannot acquire the maximum value. Second, attaching message digests or signatures to sensory samples does not allow the base station to validate data authenticity. In dangerous places, WSNs must be energy-efficient and private. In this research, we present a data aggregation method known as Energy-Efficient and Privacy-Preserving (E2P2). E2P2 data aggregation utilizes less energy and yields more accurate results. Private data aggregation with increased accuracy and hybrid encryption is presented in this research. The goal is to reduce data transmission and energy use collisions and offset collision-induced loss. Extensive simulations compare E2P2 to earlier approaches. Experimental results show that E2P2 outperforms other algorithms. Good exactness, complexity, and safety are demonstrated by theoretical and simulation results.

The design and development of the Super-X device data acquisition and monitoring system

To meet the construction requirements of future fusion reactors and other large-scale superconducting magnets, China is building an advanced superconducting conductor experimental device (Super-X) to evaluate the performance of future superconducting components. To ensure the reliability, stability of the Super-X devices and the flexibility, extensibility of the experimental tests, a specialized data acquisition and monitoring system has been designed and developed, which comprises three main components: data acquisition, transmission, and storage. For the data acquisition, a driver plug-in dynamic loading mechanism is employed to shield the diversification of data acquisition devices, thereby providing a unified data interface for upper-level operations. In terms of data transmission, the publish/subscribe (pub/sub) mode of ZeroMQ was utilized to meet the high-speed sampling demands (10 kHz sampling rate) of data transmission. In addition, a data distribution service improvement strategy that employs multiple workers is proposed to enhance subscriber reception capacity and reduce data transmission loss. As for data storage, a combination of short-term and long-term databases is used to improve the efficiency of data queries. The results of testing indicate that the system can satisfy the operational and experimental requirements of the device.