Modeling Execution Time and Energy Consumption in Sensor Node Simulation

He received his Masters Degree from RWTH Aachen in 2007. His studies were funded by " DAAD-Siemens scholarship Asia 21 st century ". Hamad started his PhD in the March of 2008 under the PhD scholarship program of DAAD. His research interests include wireless sensor networks mainly focusing on modeling & simulation, protocol design, and link level optimizations. He is the lead developer of TimeTOSSIM – a time accurate extension of TOSSIM simulator. where he graduated in 2003 with honors. His studies were funded by a Fulbright and a Direct Exchange Scholarship. Olaf received his undergraduate degree from the University of Kiel. His research interests include protocol and systems engineering as well as network modeling and simulation with a focus on embedded systems. His research activities are focused on protocol and systems engineering, modeling and network simulation, Peer-to-Peer systems, sensor networks, and operating system issues of networking. Klaus actively participates in the IETF; recently RFC 3662 and RFC 3754 were published. ABSTRACT Constrained energy and computational resources of sensor nodes are the two most critical limitations of sensor networks. Moreover, sensor nodes are envisioned to be deployed at inaccessible, remote and harsh physical environments where exchanging node's power supply is not feasible. Therefore, it is essential for sensor network developers to thoroughly evaluate and fine-tune their applications from energy and timing perspectives. As incorrect energy estimates could result in underperformance or possible breakdown of real sensor deployments before fulfilling the desired operation. Modeling energy-states of each hardware device and the time duration it spends in each state is the basic requirement for accurate energy 1 prediction in discrete event based simulations. In this article we present the architecture of our power related extensions to TimeTOSSIM – a TinyOS based simulation environment for sensor network evaluation. It employs fine grained, automated instrumentation of simulation models with cycle counts derived from application binaries to enable time accurate simulations. By instrumenting the simulation models with timing information we can capture the duty-cycle and energy-state of each hardware component. As a result, the energy consumption of each component of a sensor node can be computed. The presented approach achieves highly precise time accuracy when compared to emulation.