Assured Wireless Networking: Peer-Based Validation via Spectral Clustering

摘要

The report documents the development and test results of mathematical models and algorithms for wireless computing, sensing, and communications systems. The first contribution of this research is a novel spectral clustering method able to perform grouping by examining just the signs in leading eigenvectors of the input data. This method greatly simplifies spectral clustering, while improving the speed and robustness of the clustering process. The second contribution developed a spectral-based method for validating sensor nodes via clustering of sensors based on their measurement data. With this peer validation method, the impracticality of bringing calibration instruments to the field is overcome. This allows for easy sensor validation procedures to be conducted on the spot.