Welcome to Coventry University website

I'm a final year Computer Science student at Coventry University. I became involved with the Cogent Computing Applied Research Centre when I took part in a 'Summer of Research' program funded by the Nuffield Foundation.

My work involved the implementation of a number of routing algorithms on the Cogent in-house WSN simulator SenSOR.

For my final year project I am working on streamlining the development pipeline for SenSOR by integrating the simulator and the Gumstix hardware platform so that the same environment can be used for testing both in simulation and on the actual hardware.

This work has involved not only making small changes to the core SenSOR code but also extending the current capabilities of the application.

The SENsor SimulatOR project (SenSOR) was aimed at building a configurable software simulator for cogent microsensor networks.

Dan's work over the summer sought to not only improve on existing components but intensive work was carried out on implementing a number of popular routing algorithms as well.

1. Changed main.py so that there is more comprehensive starting framework for new projects.
2. Modified Simulation.py So that it is possible to change the Broadcast distance when creating a simulation.
3. Created DebugWin a version of Johns output window where it is possible to bind events to buttons. Allowing IMHO More structured debugging
4. Added the option to "Multi-cast".

The change to simulation allows the user to specify the broadcast range of a sensor. Allowing the simulation of more widespread WSN within a smaller area. RF_Radius is now local to the Ether object so we can change it at any time within the simulation using Ether.RF_Radius = x. By not changing the constructor for the ether we have retained backwards compatibility with any previous demos.

Although most wireless communication is by broadcast I have implemented a multi-cast method to allow data to be to specific nodes. I Feel this has several advantages.

1. Synchronisation. If you send multiple individual messages via the message queue the timings can be off, With the implementations of the routing algorithms relying on time of receiving to calculate the best path this can cause “Interesting” Route choices
2. Broadcasting to multiple sensors can clutter the graph. Using a multi-cast only draws arcs between the nodes that are supposed to be communicating. This has the advantage of making it clearer where information is being sent (My primary reason for doing this, due to debugging perils).
3. It makes code slightly tidier as there is no need to check if a directed broadcast is for a specific node.

The Rumor Routing Algorithm

The Rumor Routing algorithm is an event based routing protocol. It attempts to create paths throughout the network leading to each event. As queries are generated they can be forwarded randomly until they encounter a path leading to the event Instead of being flooded through the network. The paper referenced below gives full details on the theory behind rumor routing.

You can contact Dan at: goldsmid@coventry.ac.uk