Use of Forth to Enable Distributed Processing on Wireless Sensor Networks

Wireless Sensor Networks (WSNs) are composed of tiny sensor nodes able to monitor environmental conditions. Existing applications for WSNs usually adopt a centralized approach that exploit sensor nodes just for sensing, while data processing takes place on more powerful base stations. This can be considered a consequence of the common WSN programming practice that proves too rigid to support development based on distributed processing. In fact, local processing of complex data, such as symbolic information and rules, is an under explored aspect. The adoption of high level interpreters above general purpose operating systems is often unpractical since it implies the saturation of the available resources. In this paper, we detail the implementation of an alternative Forth-based approach that implements a minimal but extensible operating system featuring common WSN functionalities as well as advanced skills such as symbolic distributed processing. We show the definition of words and syntactic constructs that enable collaborative processing on WSNs and ease the development of complex applications even on resource constrained WSN nodes. To this purpose, our approach is based on an abstract mechanism enabling nodes to exchange directly Forth code. Cooperative behaviors, introducing dynamic computation into the network, are thus easily implemented, as we show in a few applicative examples. Moreover, using the same mechanism, remote nodes can be effortlessly reprogrammed even after their deployment. Finally, we show how our approach proves to be feasible and advantageous through a comparison, in terms of memory usage, with relevant interpreter-based software platforms for WSNs.