The main goals of the CALLAS (Calculi and Languages for Sensor Networks) research project are: to contribute towards a fundamental
understanding of wireless sensor networks by developing formal tools based on process calculi; to provide adequate programming architectures
for dynamically configurable sensor nodes; and to validate the proposed designs in real-life prototype sensor applications.
Wireless sensor networks, made of tiny, low-cost devices capable of sensing the physical world and communicating over radio links, are significantly different from classical wireless networks such as GSM or wireless LANs: (a) the design of a sensor network is strongly driven by its particular application, (b) sensor nodes are highly constrained in terms of power consumption and computational resources (CPU, memory), and (c) large-scale sensor applications require self-configuration and distributed software updates without human intervention.
Previous work on fundamental aspects of wireless sensor networks has mostly focused on communication-oriented models, in which the sensor
nodes are assumed to store and process the data, coordinate their transmissions, organize the routing of messages within the network, and relay the data to a remote receiver. Although some of these models provide useful insights (e.g. into the connectivity characteristics or the overall power efficiency of sensor networks) there is a strong need for formal methods that capture the inherent processing and memory constraints, and
illuminate the massively parallel nature of in-network computation.
Our take is to develop process calculi for sensor networks building on formalisms for concurrent distributed systems and mobile code (most
notably, pi-calculus). The envisioned tools will enable us to (a) prove the correctness and feasibility of different data gathering and communication
protocols (e.g. by exploiting behavioral equivalence between processes), and (b) carry out formal comparisons of different schemes by quantifying
the minimum amount of required resources.
Based on the results of the first task we will attack another fundamental problem in sensor networks: the need for sensor programming languages
with self-configuration and distributed update capabilities. We propose to develop specifically tailored high level idioms for sensor networks that
capitalize on notions such as mobile code and self-maintenance.
Finally, we will implement and test the developed programming tools on a sensor networking prototype. Real-life experiments will allow us to
better understand sensor networking constraints and further improve our designs. We believe that our interdisciplinary team of computer scientists and electrical engineers is well equipped to contribute both in theory and in practice to the advancement of sensor network programming.
|Start Date: 01-03-2008|
|End Date: 01-04-2011|
|Team: Gerhard Maierbacher, Joao Francisco Cordeiro de Oliveira Barros|
|Partners: FCUP, Prof. Luís Lopes (PI)|
|Local Coordinator: Joao Francisco Cordeiro de Oliveira Barros|
|Links: Internal Page|