V.N. Alexandrov et al. (Eds.): ICCS 2006, Part II, LNCS 3992, pp. 1073 – 1080, 2006. © Springer-Verlag Berlin Heidelberg 2006 On the Effect of Heterogeneous Traffic Sources on the Network Availability for Wireless Sensor Grids Ali Hammad Akbar 1 , Ki-Hyung Kim 1, *, Shaokai Yu 2 , and Won-Sik Yoon 3 1 Graduate School of Information and Communication 2 Ajou University (Invited Faculty Programme) 3 School of Electrical Engineering Ajou University, Suwon, Korea, 443-749 {hammad, kkim86, wsyoon}@ajou.ac.kr, yshaokai@yahoo.com Abstract. Wireless sensor grids are poised for applications where pre- deployment information is highly available to optimize communication and computation costs. Using two widely known grid traversal schemes, we critique the effect of multiple services being rendered by a sensor network onto its availability. We derive an interesting result using Markovian model to appreci- ate the effect of load heterogeneity onto node and network lifetimes. Our mathematical model also analyzes synchronous sleep for its effect on energy conservation. We also propose a epicenter regulated, asynchronous sleep scheme which allows nodes to conserve energy yet relay data towards the sink effectively. The performance results show that the proposed scheme prolongs network availability in sensor grids. 1 Introduction Wireless sensor networks are unique in the sense that they are supposed to work unat- tended. For their longevity, network designers propose schemes that reduce the com- putation and communication model. The fact that sensor networks have a growing application in telematics, home control, and ubiquitous services is a relief to sensor network community, wherein protocol designs are simple yet efficient. In wireless sensor grids, pre-deployment information is much useful to the design- ers. They can trim heavy and cumbersome protocols of general purpose sensor net- works to have simpler equivalents by exploiting location awareness. One such kind of information is the type of traffic that flows over the sensor grid [1]. For non-real time services, traffic flow is highly adjustable. Vice versa, real time services’ traffic flows have high variance due to greedy approach and resulting con- tention. Network availability as identified in [2] provides researchers with a baseline to diagnose and regulate node and network parameters for enhancing operational lifetime. In [3], the authors suggest a scheme for optimal transmission range distribu- tion to balance out power consumption across sensor nodes as a linear programming * Corresponding author.