Modeling and Optimization of Parallel and Distributed Embedded Systems

Chapter 9Online Algorithms for Dynamic Optimization of Embedded Wireless Sensor Networks^*

An embedded wireless sensor network (EWSN) typically consists of a set of spatially distributed embedded sensor nodes that wirelessly communicate with each other to collectively accomplish an application-specific task. Due to technological advancements in wireless communications and embedded systems, there exists a plethora of EWSN applications, including security/defense systems, industrial automation, health care, and logistics.

Given the wide range of EWSN applications, an application designer is left with the challenging task of designing an EWSN while taking into consideration application requirements (lifetime, throughput, reliability, etc.). Moreover, these application requirements are affected by environmental stimuli (e.g., poor wireless channel conditions may necessitate increased transmission power) and can change over time as operational situations evolve (e.g., unexpected winds fuel a dying forest fire). Since commercial off-the-shelf (COTS) embedded sensor nodes have limited resources (i.e., battery lifetime, processing power), delicate design and trade-off considerations are necessary to meet often competing application requirements (e.g., high processing requirements with long lifetime requirements).

In order to meet a wide range of application requirements, COTS embedded sensor nodes are generically designed, but however, tunable parameters (e.g., processor voltage, processor ...