Analysis of Latency and Related Tradeoffs in Distribute Sensor Networks

Consider a sensor network model in which nodes jointly detect an event and report related information to the "sink node". In this paper, we address the problem of finding analytical estimates for the average delay (called latency) incurred in reporting the sensed event to the sink. We consider an application model wherein events occur as a Poisson process in time, with the location of the event being uniform throughout the deployment area. We further assume that sensor nodes go to sleep and wake up independently, the sleep - wake up transition occurring as a Markov process. For such a model, we obtain analytical bounds for the average latency for reporting an event, as a function of the location of the event with respect to the sink, as well as a function of several network parameters such as network size, radio range, data rate, duty cycle ratio and wireless bandwidth. We provide detailed analysis and simulation results for the case when nodes are deployed in a regular gridlike fashion, and a detailed sketch of similar analysis for uniform random deployment.