The Wireless Philadelphia Network (WPN) is a metropolitan?area network (MAN) consisting of thousands of Tropos 5210 wireless mesh routers distributed across the entire city of Philadelphia and connected by a fiber backbone. This project is employing this network as a testbed to investigate three diverse security challenges facing any large-scale wireless network servicing a heterogeneous population. The first challenge is in efficient network anomaly detection algorithms, and the proposed solution is to investigate the efficacy of both compressive sampling and distributed source coding based approaches in reducing the amount of data that must be transmitted to the anomaly detector. The second challenge is physical layer security in wireless networks, and the proposed solution is to use physical layer based encryption algorithms and user authentication. The third challenge is anomaly detection at the application layer, in particular for web servers, and the proposed solution is to develop software sensors on the hardware, operating system, virtual machine, and application server, and develop rules for identifying possible anomalies using these metrics. Besides the intellectual merit of these challenges, the project has several broader impacts. First, low-income residents gain Internet access through integration with the Freedom Rings Partnership. Second, students participate in community service based engineering design projects. Finally, curricular enhancements and the recruitment of women and minority graduate students improve the educational and diversity missions at our university.
