The research in this proposal outlines a robust system of a mobile sensor network and develops statistical algorithms and models to provide consistent, pervasive and dynamic surveillance for nuclear or biological materials in major cities. Specifically, the project proposes a novel design of sensor network, in which nuclear sensors and Global Position System (GPS) tracking devices are installed on a large number of vehicles such as taxicabs and police vehicles. Real time information from this network is processed at a central surveillance center, where mathematical and statistical analyses are performed. The proposed statistical approaches include multiple spatial cluster detection method using latent models and state-space model for real-time detection and tracking of possibly moving nuclear sources using sequential Monte Carlo method. The methods are general and flexible and can be used in other settings involving a massive network of sensors.
Threats to national and homeland security have become more dynamic and complex in the past decade due to global terrorism, increased opposition to U.S. interests, and increased access by adversaries to sophisticated technologies and materials. Among the threats, nuclear attacks is one of the most devastating acts, with severe losses of human lives as well as long term and large scale damages to infrastructures. As a result, there have been growing concerns regarding the prospect of detonating nuclear materials or dirty bombs in the populous metropolitan areas. It becomes more and more vital to have sophisticated nuclear surveillance and detection systems deployed in major cities in the U.S. to protect infrastructures and human lives. The proposed research on surveillance systems of a mobile sensor network can have broad impacts in advancing our detection capabilities of terrorist threats hence improving homeland security. During the course of the proposed project, activities related to education and training of graduate and undergraduate students will be actively engaged, in preparation for their possible careers in this area.
Threats to national and homeland security have become more dynamic and complex in the past decade, and it becomes more and more vital to have sophisticated surveillance and detection systems deployed in major cities in the U.S. to protect infrastructures and human lives. The major goal of this project is to outline a robust system of a mobile sensor network and develop statistical algorithms and models to provide surveillance and detection for nuclear or biological materials in major cities. Specifically, we developed two sets of statistical models and algorithms for consistent, pervasive and dynamic surveillance -- one uses spatial latent cluster modeling approach and the other uses state-space modeling approach. Statistical models and inference are the tools related to the developments of the project. In addition to the major goal, the PIs have also conducted research and made good contributions in statistical models and parametric and nonparametric inferences. Furthermore, this research grant has also supported in part the PIs’ research to develop a new statistical inference tool known as confidence distribution, which addresses a number of new and important research questions that are highly relevant to the foundation of statistical inference and applications. The PIs’ other research activities include developments of models and optimization algorithms for applications in defense and homeland security, and statistical applications in finance, biological and medical sciences. The PIs have published forty articles, many of which are in top statistical journals. The PIs have also been invited to present their research results in major statistics and computer sciences conferences, as well as statistics, mathematics, computer sciences departments worldwide. The accomplished research products/activities have solved the specific set of problems set forth in this grant, and also have provided novel developments in statistical models, inference and applications. They have a board impact on statistical developments and applications in the fields of defense and homeland security, finance, education and biological and medical sciences. During the funding period, 12 Ph.D. students (five female) graduated and one postdoctoral researcher (female) was trained by the PIs. Four undergraduate students (one African American/two female students) from an NSF sponsored REU program at Rutgers DIMACS center (Center for Discrete Mathematics and Theoretical Computer Science) have also been involved in some parts of the research. The students have gained valuable research experience and made significant contributions for a variety of research projects that are related (directly or indirectly) to the grants. The involvement of graduate students and post-doc has resulted in a number of research articles and software packages. Through the involvement, students have acquired hands-on experience with real life problems. Such training is essential for them to become statisticians capable of collaborating effectively with non-statistician researchers in areas such as defense, finance and economics, medical field, engineering and industry.