DNNOff Offloading DNN-based Intelligent IoT Applications in Mobile Edge Computing
Abstract : Deep neural network (DNN) has become increasingly popular in industrial IoT scenarios. Due to high demands on computational capability, it is hard for DNN-based applications to directly run on intelligent end devices with limited resources. Computation offloading technology offers a feasible solution by offloading some computation-intensive tasks to the cloud or edges. Supporting such capability is not easy due to two aspects: (1) Adaptability: offloading should be dynamically occur among computation nodes. (2) Effectiveness: it needs to be determined which parts are worth offloading. This paper proposed a novel approach, called DNNOff. For a given DNN-based application, DNNOff first rewrites the source code to implement a special program structure supporting on-demand offloading, and at runtime, automatically determines the offloading scheme. We evaluated DNNOff on a real-world intelligent application, with three DNN models. Our results show that, compared with other approaches, DNNOff saves response time by 12.4%-66.6% on average.
? The edge server is closer to the device than the cloud server, so it has greater bandwidth and response time.
? However, most existing computation offloading frameworks for blockchain mining services have ignored user privacy.
? There are still some shortcomings in existing deep learning methods, e.g., the slow learning speed and the failure of the original network parameters when the environment changes.
? To tackle these challenges, we propose a Deep Meta Reinforcement Learning-based Offloading (DMRO) algorithm, which combines multiple parallel DNNs with Q-learning to make fine-grained offloading decisions.
? In light of this situation, the strategy of computation offloading has been adopted to solve this problem.
? A promising technique to solve this problem is to offload computationally intensive tasks to nearby servers with more abundant resources, which is called computation offloading or nomadic services.
? The multi-user computation offloading problem in a multi-channel wireless interference environment was studied in, and the game theory method was used to implement effective channel allocation in a distributed manner.
? The two-layer optimization method was used to decouple the original NP-hard problem into a low-level problem to seek power and sub-carrier allocation and upper-layer task offloading.
• The proposed work considers all the important parameters in the cost function and generates a comprehensive training dataset with high computation and complexity.
• The proposed work considers the partitioning process in a partial offloading technique and calculates cost for each possible partitioning and offloading policy and then select the partitioning and offloading policy with minimum cost.
• The proposed cost function also considers the propagation delay, radio resources, and computing resources.
• In the proposed work, first we divide a task into n components and then using the partial offloading technique, the UE offloads some of the components to MES and some of the components are executed on UE.
? In this paper, we propose a computation offloading strategy under a scenario of multi-user and multi-mobile edge servers that considers the performance of intelligent devices and server resources.
? Computation offloading is a promising technique that can promote the lifetime and performance of smart devices by offloading local computation tasks to edge servers.
? However, the effect of server resources used in computation offloading performance was not considered.
? A task scheduling model has been proposed based on the improved auction to optimize the time requirements of the tasks and computation performance of the MEC servers.
? The improved auction algorithm proposed in this paper not only has advantages in time complexity but also improves the efficiency of virtual machines.
|