Dynamic Data Processing Middleware for Sensor Networks

Heterogeneous and resource-constrained sensors, computational and communication latencies, variable geospatial deployment and diversity of applications set challenges for sensor network middleware. RESTful architecture principles have been widely applied in middleware design. The new Computational REST architecture offers additional set of principles. In Computational REST, computations are seen as resources and interactions are conducted as computational exchanges. In this PhD thesis work, these principles are studied and elaborated further in the context of sensor network middleware. Middleware and system component prototypes are developed, evaluated and utilized by field trials in real-world settings. As a result, new knowledge is generated of ubiquitous sensor network middleware design and dynamically distributing data processing computational load in resourceconstrained sensor networks.