Joint Task Placement, Routing and Power Control for Low Power Mobile Grid Computing in Ad Hoc Network

Abstract

Nowadays, almost all mobile devices in the market have Wireless Network Interface Cards (WNICs) attached that enable them to construct Ad Hoc network and communicate flexibly with each other. Most emerging mobile applications, which are usually computation and data access intensive, have different power consumption on different devices. Hence, with aim at lower power consumption, we can apply the on-demand grid computing to mobile devices, migrating task among mobile devices or, if available, stationary computers. Unavoidably, the task migration introduces additional communication activities, which result in significant power consumption on WNIC. Most existing papers use oversimplified communication energy model, and none of them fully exploits the important features of wireless Ad Hoc network to aggressively reduce communication energy consumption. Taking a cross-layer approach, we present a general modeling framework for joint task placement, routing and power control in the context of low power mobile grid computing. The modeling framework can serve as the basis of various network aware optimizations targeting on energy consumption, energy dissipation rate, network life time, etc.. Moreover, we propose an algorithm for minimizing global energy consumption, and study how various factors, such as WNIC, device density, communication penalty, etc., affect the potential of energy saving and the communication pattern in the mobile grid computing.