Multi-agent and reinforcement learning based code offloading in mobile fog
- 1 January 2016
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 285-290
Abstract
Fog computing, which performs on network edges, is a front-end distributed computing archetype of centralized cloud computing. Mobile Fog is a special purpose computing prototype, which leverages the mobile computing to deliver seamless and latency-aware mobile services. Offloading computation in mobile Fog is challenging because of the spatiotemporal resource requirements of heterogeneous mobile devices. In this paper, we propose reinforcement learning based code offloading mechanism to ensure low-latency service delivery towards mobile service consumers. We use the distributed reinforcement learning algorithm to offload basic blocks in a decentralized fashion to deploy mobile codes on geographically distributed mobile Fogs. We simulate the proposed prototype using OMNeT++ considering fluctuated resources of mobile Fog and varied service demands of mobile users. The proposed method significantly reduces the execution time and latency of accessing mobile services while ensuring lower energy consumption of mobile devices.Keywords
This publication has 12 references indexed in Scilit:
- A Fog based system model for cooperative IoT node pairing using matching theory2015
- Toward service selection game in a heterogeneous market cloud computing2015
- The Role of Cloudlets in Hostile EnvironmentsIEEE Pervasive Computing, 2013
- Coordinated resource provisioning and maintenance scheduling in cloud data centers2013
- Fog computing and its role in the internet of things2012
- Coordinated Self-Configuration of Virtual Machines and Appliances Using a Model-Free Learning ApproachIEEE Transactions on Parallel and Distributed Systems, 2012
- ThinkAir: Dynamic resource allocation and parallel execution in the cloud for mobile code offloading2012
- CloneCloud2011
- Scavenger: Transparent development of efficient cyber foraging applications2010
- Distributed reinforcement learningRobotics and Autonomous Systems, 1995