Availability Modeling and Analysis of Autonomous In-Door WSNs Safwan Al-Omari and Weisong Shi Wayne State University {somari, weisong}@wayne.edu Abstract Availability analysis and modeling in autonomous and remotely administered systems that are composed of cheap and failure-prone components is vital to redundancy man- agement, which includes the prediction of the required number of components and the way these components are scheduled ON and OFF. Targeting the application of wireless sensor networks for the monitoring of el- der people living in their apartments, we use techniques from reliability theory to model the WSN as a κ-out-of- m system with independent components. In addition to predicting the required redundancy to meet the desired availability behavior early in the planning phase, we show that scheduling these nodes ON and OFF later on in the operational phase does indeed improve availability over the entire system lifetime. To validate our model, we design a simulator using nesC/TOSSIM. Our analytical and experimental results show that using node scheduling almost doubles the expected WSN total uptime. I. Introduction Wireless Sensor Networks (WSNs) are envisioned to operate autonomously for long periods of time without close maintenance and supervision. This makes modeling and predicting WSNs availability behavior a very impor- tant and challenging problem. This problem includes two aspects. One is about planning for the required number of nodes to meet desired availability behavior. The other is in controlling the way these nodes are turned ON and OFF (i.e., node scheduling) to improve the system’s availability. By availability we mean the probability that the system (i.e., WSN) will be available (i.e., functioning) at an arbitrary point of time over its lifetime. In contrast to avail- ability, reliability means the probability of having longer continuous and uninterrupted operation of the system. 1-4244-1455-5/07/$25.00 c 2007 IEEE. Developing such models requires a clear understanding of sensor node failure models. From the perspective of failure models, we can identify two application classes: in-door [4] and out-door [9] WSN applications. The latter tends to have a hostile deployment environment, which results in more frequent and unexpected node failures due to harsh environmental conditions [11], [12]. Whereas, the former tends to provide more controlled deployment environment, which results in fewer and more expected sensor node failures. Therefore, adopting the usage-based failure model, in which the probability of failure depends on the time a node spends in the ON mode, is more rea- sonable. Despite the controlled deployment environment in in-door WSNs, sensor nodes are still failure-prone due to their resource limitation and cheap cost. Therefore, WSN deployments are envisioned to involve high degree of node redundancy [2], [7], [11] to overcome these limitations. In topology management protocols [2], [3], [4], [10], node redundancy and scheduling has been used basically to extend the WSN lifetime beyond the lifetime of a single sensor node. Nodes are redundantly deployed so that multiple sensors are able to perform the same function. Some nodes are turned OFF and save energy, while others stay ON and perform the function. Two functions were con- sidered, connectivity and coverage. Our work complements their work by emphasizing availability as a new require- ment on the node redundancy and scheduling algorithms. As topology (either connectivity or coverage) is not our concern, we prefer to use redundancy management instead of topology management in this paper. A. Motivation of Node Scheduling Note that turning a node OFF in the usage-based failure model prevents node failures. Hence, node scheduling does indeed affect WSN availability behavior. In Fig. 1, we show an interesting result on how node scheduling can affect WSN availability. The x-axis represents time, and the y-axis represents system availability. In the queue scheduling scheme, the m sensor nodes are divided in