A Model-Driven Approach for Data Collection in Sensor Networks

G. Singh and S. Pujar (USA)


Sensor networks, Data Distribution, Automated Software Development.


Many applications such as those for surveillance and target detection rely on a network of sensors to collect real-time data. These sensor applications may have to operate semi-autonomously under a diverse set of operating conditions. Often, several alternative algorithms are available for each service required by an application, and the performance of these algorithms may depend on several factors such as network topology, placement of sensor nodes, and the application structure. As a result, designers are often faced with the problem of identifying the best set of algorithms to use for an application. We study this problem in the context of rate-based data propagation. This problem involves constructing a data propagation tree to efficiently disseminate data from a sensor node at the required rate to each of its consumers. We identify a number of operating conditions that can influence the efficiency of data collection trees. We propose a set of algorithms to construct data propagation trees, and perform experimental evaluation to determine the operating conditions under which each algorithm outperforms the others. Based on these evaluations, we provide heuristics to choose the most appropriate algorithm for the given operating conditions.

Important Links:

Go Back