scholarly journals Mobility-Included DNN Partition Offloading from Mobile Devices to Edge Clouds

Sensors ◽  
2021 ◽  
Vol 21 (1) ◽  
pp. 229
Author(s):  
Xianzhong Tian ◽  
Juan Zhu ◽  
Ting Xu ◽  
Yanjun Li
Keyword(s):  
Neural Networks ◽  
Energy Consumption ◽  
Mobile Devices ◽  
Wireless Network ◽  
Deep Neural Networks ◽  
Mobile User ◽  
Computation Offloading ◽  
Long Latency ◽  
Total Latency ◽  
And Performance

The latest results in Deep Neural Networks (DNNs) have greatly improved the accuracy and performance of a variety of intelligent applications. However, running such computation-intensive DNN-based applications on resource-constrained mobile devices definitely leads to long latency and huge energy consumption. The traditional way is performing DNNs in the central cloud, but it requires significant amounts of data to be transferred to the cloud over the wireless network and also results in long latency. To solve this problem, offloading partial DNN computation to edge clouds has been proposed, to realize the collaborative execution between mobile devices and edge clouds. In addition, the mobility of mobile devices is easily to cause the computation offloading failure. In this paper, we develop a mobility-included DNN partition offloading algorithm (MDPO) to adapt to user’s mobility. The objective of MDPO is minimizing the total latency of completing a DNN job when the mobile user is moving. The MDPO algorithm is suitable for both DNNs with chain topology and graphic topology. We evaluate the performance of our proposed MDPO compared to local-only execution and edge-only execution, experiments show that MDPO significantly reduces the total latency and improves the performance of DNN, and MDPO can adjust well to different network conditions.

scholarly journals MB-CNN: Memristive Binary Convolutional Neural Networks for Embedded Mobile Devices

2018 ◽  
Vol 8 (4) ◽  
pp. 38 ◽  
Author(s):  
Arjun Pal Chowdhury ◽  
Pranav Kulkarni ◽  
Mahdi Nazm Bojnordi
Keyword(s):  
Neural Networks ◽  
Mobile Devices ◽  
Convolutional Neural Networks ◽  
Mobile User ◽  
Power Performance ◽  
Performance Requirements ◽  
System Energy ◽  
And Performance ◽  
The Cost

Applications of neural networks have gained significant importance in embedded mobile devices and Internet of Things (IoT) nodes. In particular, convolutional neural networks have emerged as one of the most powerful techniques in computer vision, speech recognition, and AI applications that can improve the mobile user experience. However, satisfying all power and performance requirements of such low power devices is a significant challenge. Recent work has shown that binarizing a neural network can significantly improve the memory requirements of mobile devices at the cost of minor loss in accuracy. This paper proposes MB-CNN, a memristive accelerator for binary convolutional neural networks that perform XNOR convolution in-situ novel 2R memristive data blocks to improve power, performance, and memory requirements of embedded mobile devices. The proposed accelerator achieves at least 13.26 × , 5.91 × , and 3.18 × improvements in the system energy efficiency (computed by energy × delay) over the state-of-the-art software, GPU, and PIM architectures, respectively. The solution architecture which integrates CPU, GPU and MB-CNN outperforms every other configuration in terms of system energy and execution time.

scholarly journals A Computational Offloading Method for Edge Server Computing and Resource Allocation Management

2021 ◽  
Vol 2021 ◽  
pp. 1-11
Author(s):  
Muna Al-Razgan ◽  
Taha Alfakih ◽  
Mohammad Mehedi Hassan
Keyword(s):  
Resource Allocation ◽  
Response Time ◽  
Real Time ◽  
Mobile Devices ◽  
Mapping Function ◽  
Mobile User ◽  
Computation Offloading ◽  
Efficient Resource ◽  
And Performance ◽  
Cloud Servers

The emerging technology of mobile cloud is introduced to overcome the constraints of mobile devices. We can achieve that by offloading resource intensive applications to remote cloud-based data centers. For the remote computing solution, mobile devices (MDs) experience higher response time and delay of the network, which negatively affects the real-time mobile user applications. In this study, we proposed a model to evaluate the efficiency of the close-end network computation offloading in MEC. This model helps in choosing the adjacent edge server from the surrounding edge servers. This helps to minimize the latency and increase the response time. To do so, we use a decision rule based Heuristic Virtual Value (HVV). The HVV is a mapping function based on the features of the edge server like the workload and performance. Furthermore, we propose availability of a virtual machine resource algorithm (AVM) based on the availability of VM in edge cloud servers for efficient resource allocation and task scheduling. The results of experiment simulation show that the proposed model can meet the response time requirements of different real-time services, improve the performance, and minimize the consumption of MD energy and the resource utilization.

scholarly journals Flexible computation offloading in a fuzzy-based mobile edge orchestrator for IoT applications

2020 ◽  
Vol 9 (1) ◽  
Author(s):  
VanDung Nguyen ◽  
Tran Trong Khanh ◽  
Tri D. T. Nguyen ◽  
Choong Seon Hong ◽  
Eui-Nam Huh
Keyword(s):  
Cloud Computing ◽  
Augmented Reality ◽  
Mobile Devices ◽  
Computing System ◽  
Mobile User ◽  
Computation Offloading ◽  
Context Aware ◽  
Computing Systems ◽  
Iot Applications ◽  
Service Times

AbstractIn the Internet of Things (IoT) era, the capacity-limited Internet and uncontrollable service delays for various new applications, such as video streaming analysis and augmented reality, are challenges. Cloud computing systems, also known as a solution that offloads energy-consuming computation of IoT applications to a cloud server, cannot meet the delay-sensitive and context-aware service requirements. To address this issue, an edge computing system provides timely and context-aware services by bringing the computations and storage closer to the user. The dynamic flow of requests that can be efficiently processed is a significant challenge for edge and cloud computing systems. To improve the performance of IoT systems, the mobile edge orchestrator (MEO), which is an application placement controller, was designed by integrating end mobile devices with edge and cloud computing systems. In this paper, we propose a flexible computation offloading method in a fuzzy-based MEO for IoT applications in order to improve the efficiency in computational resource management. Considering the network, computation resources, and task requirements, a fuzzy-based MEO allows edge workload orchestration actions to decide whether to offload a mobile user to local edge, neighboring edge, or cloud servers. Additionally, increasing packet sizes will affect the failed-task ratio when the number of mobile devices increases. To reduce failed tasks because of transmission collisions and to improve service times for time-critical tasks, we define a new input crisp value, and a new output decision for a fuzzy-based MEO. Using the EdgeCloudSim simulator, we evaluate our proposal with four benchmark algorithms in augmented reality, healthcare, compute-intensive, and infotainment applications. Simulation results show that our proposal provides better results in terms of WLAN delay, service times, the number of failed tasks, and VM utilization.

Joint Optimization Offloading Strategy of Execution Time and Energy Consumption of Mobile Edge Computing

2021 ◽  
Vol 18 (5) ◽  
Author(s):  
Qingzhu Wang ◽  
Xiaoyun Cui
Keyword(s):  
Energy Consumption ◽  
Mobile Devices ◽  
Execution Time ◽  
Optimal Solution ◽  
Computation Offloading ◽  
Joint Optimization ◽  
Random Strategy ◽  
Fitness Value ◽  
Cloud Server ◽  
Time And Energy

As mobile devices become more and more powerful, applications generate a large number of computing tasks, and mobile devices themselves cannot meet the needs of users. This article proposes a computation offloading model in which execution units including mobile devices, edge server, and cloud server. Previous studies on joint optimization only considered tasks execution time and the energy consumption of mobile devices, and ignored the energy consumption of edge and cloud server. However, edge server and cloud server energy consumption have a significant impact on the final offloading decision. This paper comprehensively considers execution time and energy consumption of three execution units, and formulates task offloading decision as a single-objective optimization problem. Genetic algorithm with elitism preservation and random strategy is adopted to obtain optimal solution of the problem. At last, simulation experiments show that the proposed computation offloading model has lower fitness value compared with other computation offloading models.

Towards Stealing Deep Neural Networks on Mobile Devices

2021 ◽  
pp. 495-508
Author(s):  
Shashank Reddy Danda ◽  
Xiaoyong Yuan ◽  
Bo Chen
Keyword(s):  
Neural Networks ◽  
Mobile Devices ◽  
Deep Neural Networks

scholarly journals SNR: S queezing N umerical R ange Defuses Bit Error Vulnerability Surface in Deep Neural Networks

2021 ◽  
Vol 20 (5s) ◽  
pp. 1-25
Author(s):  
Elbruz Ozen ◽  
Alex Orailoglu
Keyword(s):  
Neural Networks ◽  
Deep Learning ◽  
Deep Neural Networks ◽  
Low Cost ◽  
Error Resilience ◽  
Error Rates ◽  
Training Methods ◽  
Performance Requirements ◽  
Proper Design ◽  
And Performance

As deep learning algorithms are widely adopted, an increasing number of them are positioned in embedded application domains with strict reliability constraints. The expenditure of significant resources to satisfy performance requirements in deep neural network accelerators has thinned out the margins for delivering safety in embedded deep learning applications, thus precluding the adoption of conventional fault tolerance methods. The potential of exploiting the inherent resilience characteristics of deep neural networks remains though unexplored, offering a promising low-cost path towards safety in embedded deep learning applications. This work demonstrates the possibility of such exploitation by juxtaposing the reduction of the vulnerability surface through the proper design of the quantization schemes with shaping the parameter distributions at each layer through the guidance offered by appropriate training methods, thus delivering deep neural networks of high resilience merely through algorithmic modifications. Unequaled error resilience characteristics can be thus injected into safety-critical deep learning applications to tolerate bit error rates of up to at absolutely zero hardware, energy, and performance costs while improving the error-free model accuracy even further.

Sign in / Sign up

Export Citation Format

Share Document