Adaptive Computational Workload Offloading Method for Web Applications

Abstract

Nowadays, cloud computing has become a common computing infrastructure. As the computing paradigm has been shifted to cloud computing, devices can utilize computing resource any-where/any-time/any-device. Many research papers in mobile cloud called ‘cloud offloading’ which migrates a device’s workload to a server or to other devices have been proposed. However, previous cloud offloading methods are mainly focusing on the cloud offloading between a device and a server. Furthermore, these proposed methods have rarely commonly used because the proposed methods were very complex - difficulty of partitioning application tasks and maintaining execution status sync between a device and a server in the cloud. In this paper, I proposed the adaptive framework for cloud offloading based on the web application standard - HTML5 specification - for web applications based on flexible resource in servers as well as in devices. In HTML5 specification, there is the method for the parallel execution of the task named ‘Web Worker’ and the method for the communication between a device and a server named ‘Web Socket’. Utilizing the property of this specification, I proposed a seamless method to do the cloud offloading for parallelized tasks of the web applications among devices as well as between a device and a server.. Based on proposed method, a device can seamlessly migrate a part of web application workload with the Web Worker to other resource owners - devices and servers - with a little modification of web applications. As a result, I can successfully build the environment where a device which has a HTML5 browser such as a mobile phone and a smart TV can share the workload among devices and servers in various situations – out-of-battery, good network connection, more powerful computing needs.