As the number and proportion of compute-intensive tasks on mobile devices increase, so does the amount of energy required to process them. Edge computing technologies provide a solution to enhance the capacity of mobile devices by offloading those tasks to edge servers. Based on dynamic voltage and frequency scaling (DVFS) technology, this paper proposed a new energy saving offloading strategy for mobile devices. On the premise that the task completion time meets the task deadline, the energy consumption problem of mobile devices is formulated as a minimum problem and solved by the genetic algorithm. In this strategy, priority is allocated to each task of workflow, and corresponding offloading resources are allocated according to the result of classification. DVFS technology is applied to the CPU of the terminal device to further reduce energy consumption while the task-resource mapping is performed. Experimental results show that for the same workflow and the same offloading resources and communication environment, in meet the delay constraints of tasks, the proposed algorithm is compared with the existing fine-grained task migration energy-saving algorithm based on genetic algorithm and the task scheduling algorithm based on energy consumption perception, can effectively improve the high energy consumption of mobile devices.

How to ensure the reliability of mixed communication between vehicle to vehicle and vehicle to road, and how to ensure seamless handover during communication in the Internet of Vehicles, is a research hotspot in 5G cellular Internet of Vehicles. This paper proposes a 5G cellular Internet of Vehicles communication handover strategy model for V2V (Vehicle to Vehicle) and V2I (Vehicle to Infrastructure) scenarios, which combines fuzzy logic and user satisfaction for obtaining quality of service, and can make intelligent decisions on handover trigger timing and handover targets based on various parameters in the wireless network, so that the vehicle can handover to the best service target in the process of vehicle terminal handover. The experimental study shows that the proposed method outperforms other common methods in terms of handover metrics such as handover trigger rate, handover times and quality of service compared to existing studies.

In the cloud-edge collaborative computing environment, how to allocate appropriate edge computing and cloud computing resources for tasks according to their quality of service requirements is the key to ensuring the effectiveness of cloud-edge collaboration. Aiming at the decision-making problem of offloading tasks with different deadlines in the cloud-edge collaborative system, a strategy of prioritizing tasks based on deadlines and then allocating cloud, edge computing resources, and communication resources based on priority is proposed. The theoretical analysis model and heuristic algorithm are presented, and then the performance of the proposed theoretical model and the algorithm is verified and analyzed through simulation experiments.

The emergence of edge computing has greatly improved the computing and storage capabilities of mobile terminals and IoT devices. However, most of the current research in the edge computing field focuses on computing offloading, resource allocation, and computing migration, and there is very little research on the placement of edge servers. To address the problem of edge server placement, we have proposed an edge server placement algorithm (MRT) based on the suitability evaluation of access points in the 5G ultra-dense network environment. The algorithm first uses the analytic hierarchy process (AHP) and the entropy weight method to evaluate the suitability of each access point based on characteristic indicators to determine whether the access point is suitable for placing edge servers. Then, according to the suitability evaluation results, the solution space of the edge server placement is reduced, and the optimal solution for the edge server is found in the reduced solution space through the tabu search method. Sufficient experimental results show that the proposed MRT algorithm can effectively reduce the average response delay of computing tasks, and the algorithm performance is better than other typical placement algorithms.

Abstract Object detection is one of the main tasks of computer vision. Object detection algorithms usually rely on deep convolutional neural networks, which require the host device to have high computing capabilities, greatly limiting the application of object detection methods for mobile devices with limited computing capabilities, such as embedded devices. Among the current object detection algorithms, the you only look once (YOLO) series takes both speed and accuracy into consideration and is one of the most commonly used methods for object detection. In this article, TRC‐YOLO is proposed, which improves the mean average precision (mAP) and real‐time detection speed of the model while reducing the size of the model. In TRC‐YOLO, the convolution kernel of YOLO v4‐tiny is pruned and an expansive convolution layer is introduced into the residual module of the network to produce an hourglass Cross Stage Partial ResNet (CSPResNet) structure. A receptive field block (RFB) that simulates human vision is also added, increasing the receptive field of the model and strengthening the feature extraction ability of the network. In addition, the convolutional block attention module is applied, which combines spatial attention and channel attention, to enhance the effective features of the model and reduce the negative impact of noise on the model. The size of the TRC‐YOLO model is 17.8 MB, which is 5.9 MB smaller than YOLO v4‐tiny, and the model parameter is 2.983 billion floating point operations per second (BFLOP/s) (3.834 BFLOP/s less than YOLO v4‐tiny). In addition, TRC‐YOLO achieves a real‐time performance of 36.9 frames per second on a Jetson Xavier NX, and its mAP on the PASCAL VOC dataset is 66.4 (3.83 higher than YOLO v4‐tiny). In addition, the mAP of TRC‐YOLO on the MS COCO dataset is 37.7, which is 1.9 higher than that of the baseline model.

Abstract This paper introduce an efficient object detection network named Trident‐You Only Look Once (YOLO), which is designed for mobile devices with limited computing power. The new architecture is improved based on YOLO v4‐tiny. The authors redesign the network structure and propose a trident feature pyramid network (Trident‐FPN), which can improve the precision and recall of lightweight object detection. Specifically, Trident‐FPN increases the computational complexity by only a small amount of floating point operations per second (FLOPs) and obtains a multi‐scale feature map of the model, which significantly lightweight object detection performance. To enlarge the receptive field of the network with the fewest FLOPs, this paper redesign the receptive field block (RFB) and spatial pyramid pooling (SPP) layer and propose tinier cross‐stage partial RFBs and smaller cross‐stage partial SPPs. This paper present extensive experiments, and Trident‐YOLO shows strong performance compared to that of other popular models on the PASCAL VOC and MS COCO. On the MS COCO and PASCAL VOC 2007 test sets, the mean average precision (mAP) of Trident‐YOLO improved by 4.5% and 5.0%, respectively. Trident‐YOLO also reduce the network size by more than 54.4% compared to YOLO v4‐tiny. With a 23.7% FLOP reduction, the FPS is improved by 1.9 on an Nvidia Jetson Xavier NX.

Most of the existing work about device-to-device (D2D) data offtoading do not simultaneously consider the mobile scenario and the trade-off between the revenue and cost of caching data. In this paper, we consider a more comprehensive and practical scenario of D2D data offloading, and design a mobility-aware incentive mechanism to efficiently select some mobile users to pre-cache the proper contents with the objective of maximizing social welfare by jointly considering mobile users' preference similarity and the social relationship. Simulation results demonstrates that the proposed mechanism outperforms other counterparts. In addition, the proposed mechanism satisfies the nice properties of individual rationality and truthfulness.

Contention‐based carrier sense multiple access (CSMA) and contention‐free time division multiple access (TDMA) protocol are two typical access protocols of media access control in vehicular ad hoc network (VANET). They all show their unique advantages under specific conditions. TDMA has high transmission reliability, CSMA has a low transmission delay in the communication environment with low packet arrival rate. However, when the arrival rate of information packets increases rapidly, the throughput of CSMA will decrease rapidly and approach zero, which is not suitable for data transmission in the communication environment with a high arrival rate of information packets; while data transmission through a single TDMA protocol will cause high system overhead due to strict synchronous information. Aiming at the defects of the two protocols and the multi‐channel communication environment, this study proposes the optimised protocol model multi‐priority time division‐CSMA (MPTD‐CSMA). The optimised protocol model not only ensures the reliability of data communication but also reduces the transmission delay of the system. At the same time, the multi‐priority mechanism is added to increase the channel utilisation of the protocol model.

Backfilling is well known in parallel job scheduling to increase system utilization and user satisfaction over traditional non-backfilling scheduling algorithms, which allow small jobs from the back of the queue to execute before larger jobs arriving earlier, and resources could be reserved to protect the latter from starvation. This paper proposed a relaxed backfill scheduling mechanism supporting multiple reservations, and investigated its effectiveness in reducing the average waiting time and average slowdown of jobs by using simulations with real traces. Different from existing relaxed scheduling, which restrict the maximum number of reservations to one, this new mechanism can support the relaxation of multiple reservations and works efficiently in scheduling by successful avoidance of raising chain reactions in relaxing the start times of multiple already existing reservations. Experimental results suggest that although the performances of both the relax-based backfilling and the strict backfill depend on the accuracy of runtime estimates, reservation depths, traces and system load alike, the former scheduling is more flexible and generally more effective in reducing the average waiting time and average slowdown of jobs, without loss of utilization.

Advance reservations are usually adopted to guarantee the QoS of grid applications by reserving a particular resource capability over a defined time interval on local resources. Single machine scheduling is the basis of more complicated parallel machine scheduling. Assume the information (e.g., arrival time, required processing time) on queued AR and non-AR jobs is known before scheduling, it is NP-hard to allocate non-resumable non-AR jobs into available intervals left by AR jobs to minimize the makespan. This paper investigates the cases with lookahead k in which the scheduler can preview the durations of the sequent k-1 available intervals as the current available interval arrives. By transforming this deterministic scheduling problem into a variant of the standard variable-sized bin packing problem, this paper proposed four looking-ahead algorithms and investigated their performances from both of the worst case and the average case viewpoints.