Cooperating Tools for Wireless Sensor Networks

Wireless Sensor Network (WSN) simulators, testbeds, and environment simulators are indispensable tools for WSN research. Existing WSN tools were developed with different purposes without the intention of cooperation. Nevertheless, the development of a WSN technology (e.g, a middleware, a protocol, or an application) usually requires the cooperation among multiple tools. This calls upon a breaking of the incompatible barriers between any pair of tools. In this thesis, I propose the Common Input and Output integration approach (the CIO approach). This approach attempts to define a standard configuration file format containing input and output data common to WSN tools. In hope that by supporting this standard configuration file format, tools can cooperate. The WiseML [23] format defined by the WISEBED [29] project is fully in compliance with the requirement of the CIO format. Therefore, it is chosen as a concrete representation of the CIO format. To evaluate the CIO approach, I developed a bidirectional online converter for one of the WSN simulators - COOJA [42]. The converter is capable of converting between the WiseML format and COOJA's native CSC format on-the-fly. It provides not only the support for both the input and output of the WiseML format but also the functionality of converting the text format temperature log file generated by testbeds to WiseML format. While I was working on this work, a number of other international organizations were also adding WiseML supports to their own tools. We exchanged our WiseML files and observed gratifying results: (1) WSN simulators can load each other's WiseML files and reproduce the same network topologies and scenarios in their own simulation frameworks; (2) The WiseML format scenarios generated by testbeds and environment simulators can be directly inserted into a WiseML file, and take effect during simulations; (3) The time and space overheads of the converter is acceptable and proportional to the complexity of a simulation. A WiseML file of a few hundreds KB and a format converting time of a few hundreds milliseconds can meet the requirements of most of the simulations. For example, a WiseML file containing 800 motes without scenario and trace sections has a size of approximately 200 KB, and a format converting time of around 500 milliseconds