Modeling path duration time in dynamic convergecast network

Estimating communication latency is challenging in dynamic settings, e.g mobile applications where the source and/or destination of the communication can move arbitrarily. In this paper, we explore a many-to-one communication pattern, called convergecast, where mobile sensors report their sensed values to one or more servers (data sinks) in a periodic or queried manner. Such ”convergecast” communication is becoming increasingly relevant in sensor monitoring and pervasive sensing applications. An important aspect of enabling accurate collection is to ensure small delays in the collection process. The path duration, i.e time between when a path from sensor to source is established and when it gets disrupted, is a key factor that affects end-to-end delay; much of the existing work on this topic is in the context of MANETs. We propose and evaluate a probabilistic model in convergecast to capture path duration times given models of the network/transmission, the scale of the network and mobility of network elements. We validate our model through simulations and verify that the proposed model can provide a reasonable representation for end-to-end delays in the convergecast network. Pervasive sensor networks of the future are likely to encompass multiple access technologies with varying network properties and transmission characteristics. The ability to model end-to-end network properties (such as path duration) can assist in improved utilization in such pervasive multinetworks.