The recent rapid development of urbanization and Internet things (IoT) encourages more research on Smart City in which computing devices are widely distributed huge amount dynamic real-time data collected processed. Although vast volume available for extracting new living patterns making urban plans, efficient processing instant decision still key issues, especially emergency situations requesting quick response with low latency. Fog Computing, as the extension Cloud enables tasks...

Mobile Edge Computing (MEC) has become an attractive solution to enhance the computing and storage capacity of mobile devices by leveraging available resources on edge nodes. In MEC, arrivals tasks are highly dynamic hard predict precisely. It is great importance yet very challenging assign nodes with guaranteed system performance. this article, we aim optimize revenue earned each node optimally offloading We formulate revenue-driven online task (ROTO) problem, which proved be NP-hard. first...

Though the rapid globally urbanization and Internet of Things (IoT) are creating great opportunities for researchers urban planners to get significant insight our cities, efficient information extraction instant decision making still key issues in which Cloud Computing is not universal solution any more. We present an traffic speeding monitoring system using Fog only can track vehicle obtain speed real-time, but also implement multi-target tracking single target algorithm. The preliminary...

Recently, many automatic test generation techniques have been proposed, such as Randoop, Pex and jCUTE. However, usually coverage of these has around 50-60% only, due to several challenges, 1) the object mutation problem, where generators cannot create and/or modify inputs desired states; 2) constraint solving fail solve path conditions cover certain branches. By analyzing branches not covered by state-of-the-art techniques, we noticed that challenges might be so difficult for humans.

Conducting federated learning across distributed sites with In-Band Network Telemetry (INT) based data collection faces critical challenges, including control decisions of different frequencies, convergence the models being trained, and resource provisioning coupled over time. To study this problem, we formulate a non-linear mixed-integer program to optimize long-term INT overhead, cost, cost. We then design polynomial-time online algorithms solve problem only observable inputs on fly,...

Chip defect detection is a crucial aspect of the semiconductor production industry, given its significant impact on chip performance. This paper proposes lightweight neural network with dual decoding paths for LED segmentation, named LDDP-Net. Within LDDP-Net framework, receptive field MobileNetv3 backbone modified to mitigate information loss. In addition, consisting coarse path and fine-grained in parallel are developed. Specifically, former employs straightforward upsampling approach,...

Provisioning machine learning inference as a service at the mobile network edge for distributed users in an online setting faces multiple challenges, including accuracy-resource trade-off model selection, time-coupled decision distribution, and unpredictable user workload. To overcome such we firstly time-varying non-linear integer program of maximizing overall service's accuracy through dynamic instance delivery workload distribution. Afterwards, design algorithm to make fractional control...

Instead of relying on remote clouds, today's Augmented Reality (AR) applications usually send videos to nearby edge servers for analysis (such as objection detection) so optimize the user's quality experience (QoE), which is often determined by not only detection latency but also accuracy, playback fluency, etc. Therefore, many studies have been conducted help adaptively choose best video configuration, e.g., resolution and frame per second (fps), based network bandwidth further improve QoE....

Computation offloading makes sense to the interaction between users and compute-intensive applications. Current researches focused on deciding locally or remotely executing an application, but ignored specific proportion of application. A full cannot make best use client server resources. In this paper, we propose innovative reinforcement learning (RL) method solve proportional computation problem. We consider a common scenario with time-variant bandwidth heterogeneous devices, device...

The high-performance generative artificial intelligence (GAI) represents the latest evolution of computational intelligence, while blessing future 6G networks also makes edge (EI) full development potential. inevitable encounter between GAI and EI can unleash new opportunities, where GAI's pre-training based on massive computing resources large-scale unlabeled corpora provide strong foundational knowledge for EI, harness fragmented to aggregate personalized GAI. However, natural...

In the process of steel strip production, accuracy defect detection remains a challenge due to diversity types, complex backgrounds, and noise interference. To improve effectiveness surface in strips, we propose an enhanced model known as YOLOv8-BSPB. First, novel pooling layer module, SCRD, which replaces max with average pooling. This module introduces receptive field block (RFB) deformable convolutional network version 4 (DCNv4) obtain learnable offsets, allowing kernels flexibly move...

The synchronization control performance of the Fieldbus system (FCS) is an important guarantee for completion multi-axis collaborative machining tasks, and its accuracy one decisive factors quality. To improve FCS, this paper first makes a comprehensive analysis affecting in FCS. Secondly, by analyzing communication model linear Ethernet, distributed clock compensation method based on timestamps proposed to solve asynchronous problem data transmission ethernet bus topology. Then, CANopen...

Deploying deep convolutional neural network (CNN) to perform video analytics at edge poses a substantial system challenge, as running CNN inference incurs prohibitive cost in computational resources. Model partitioning, promising approach, splits CNNs and distributes them multiple devices closer proximity each other for serial inferences, however, it causes considerable cross-edge delay transmitting intermediate feature maps. To overcome this we present ResMap, new framework that...

The inference workload redistribution is a technique for evacuating requests from hot edges to idle in edge collaborative systems, thereby achieving balancing on different edges. However, with the continuous development of accelerators, resource utilization accelerators executing series often low, and when multiple parallel, it faces uncertain execution delays, response-time Service Level Objectives (SLOs), generality workloads heterogeneous systems. To address these issues, first time...

Deep neural networks (DNNs) have facilitated commendable performance in signal processing, thanks to their superior functions feature extraction and abstraction representation. However, limited computing capacity of Internet Things (IoT) devices imposes challenges support resource-intensive DNNs inference with low latency quality service requirements. Benefiting from the pervasive wireless connectivity, offloading partial fog nodes (FNs) becomes a viable solution for alleviating resource...

This paper proposes an improved sampling-based approach for spacecraft proximity operation path planning under Clohessy-Wiltshire-Hill dynamics. The proposed is based on a modified version of the FMT* (Fast Marching Tree) algorithm with safety strategy which divided into three parts: (1) incorporating relative ellipse to simplify sampling state space and avoid collision target; (2) combining internal/external ellipse-based detection algorithms hovering obstacle non-coplanar obstacle; (3)...

This paper introduced a technique for edge detection by Biomimetic pattern recognition (BPR).  The images were scanned into computer. Due to the nature of acquiring technique, acquired have lots artifacts, resulting in complicated detection.  According this, we used recognition, which is based on “matter cognition” instead classification” and rather closer function human being. Finally, experiments showed that feasible has some flexibility.