Although a great deal of research within the WSN domain has addressed key issues such as energy efficiency of communication methods, the relative immaturity of the subject area means that there is little standardisation of hardware platforms or programming methodologies. This has led to many implementations of proposed protocols having a tight coupling between the application and the hardware stack, limiting the extent to which programming best practice of code reuse and interoperability can be used. Such tightly coupled approaches to the design of WSNs make it extremely difficult to evaluate applications in any system other than that they were originally designed for. This severely limits the opportunity to evaluate interaction between algorithms operating at different levels of the overall architecture, and the comparison of these applications on different hardware platforms.
In much the same way as the development of high level languages allowed mainstream computing to flourish, we believe that the WSN community will greatly benefit from a middleware framework. Although there has been wide debate over the potential direction of a generic application framework for WSN and the usefulness of middleware approaches for wireless sensor networks discussed, little practical work in the area has been done.
This research proposes that the development of a generic middleware framework for WSN systems will make it possible to avoid monolithic approaches to WSN systems design while moving towards component-based architectures. The ultimate goal is the enabling of key functionality, such as network management and querying, facilitated via a hardware independent API (application programming interface). The work therefore focuses on the creation of a generic WSN middleware framework aimed at addressing the issues described above.
The underlying network topology has a huge effect on the choices made during system design as some topologies may not support certain functionality. Cluster based network topologies are a popular choice in WSN design, and are widely regarded as providing significant energy savings over other network topologies. I am currently undertaking a review of cluster and backbone based network infrastructures, to evaluate the appropriateness of each architecture with regard to choosing the underlying structure of the framework.
More details are available in this presentation.
Implementations
I have implemented a Python based web server for the Gumstix platform. This allows us to monitor the status of the individual Gumstix using a web interface. The interface also allows us to manage the Bluetooth network parameters of the Gumstix, modifying the scatternet configuration dynamically to create new network hierarchies. Here is a video of the web interface in action.
A Wireless Monitoring System for a Gas Turbine Engine.
A video demonstration of the system is available here: