Current technologies have made it possible for submarines to evade standard sonar detection. Finding solutions to detect intruding submarines therefore becomes important and timely. A viable approach is to deploy magnetic or acoustic sensors in close proximity of possible underwater pathways intruders may pass through. This project seeks to develop a comprehensive theoretical and practical solution to construct undersea sensor networks for intrusion detection. When sensors are randomly deployed, spatial barriers are unlikely to exist, allowing intruders to pass through the bounded 3D space undetected. This motivates the use of mobile sensors to dynamically form sensor barriers. How to minimize the energy consumed by the movement of underwater sensors is a challenging issue. This project tackles the problem via three thrusts: (1) Develop an energy-efficient approach to using mobile sensors to construct a spatial barrier in 3D space; (2) Devise near-optimal practical solutions to reduce computation and communication costs, and develop these algorithms into practical protocols; and (3) Develop simulation modules and test-beds to evaluate the proposed solutions with realistic undersea environment parameters. The project integrates concepts and techniques in auction algorithms, geometry, combinatorial optimization, underwater acoustic communications and networking, software and system development to construct analytical models and practical solutions. The research results are expected to have a substantial impact on the understanding of constructing sensing barriers in underwater environments, and will be integrated into graduate and undergraduate teaching and outreach activities. Efforts will be also proactively pursued to recruit students from under-represented groups to participate in the project.
In academic year 2013-2014, our goal is to study the efficient routing protocol in delay tolerant networks, including underwater wireless sensor networks, the virtual backbone and fault-tolerant virtual backbone in wireless sensor networks, some optimization problems in MapReduce-like systems, client assignment in client/server systems, and social influence and rumor blocking in social networks. The descovery in the study and the research experience would help participated students to prepare their career in academia and industry. The following are examples of our research results: For delay tolerant networks, we propose a routing protocol by considering consecutive forwarding quality , and conduct extensive simulations with real trace data, Simulation results show that our protocol outperforms most existing pure probability based forwarding protocols and most existing pure centrality-community based forwarding protocols. For fault-tolerant virtual backbone in wireless sensor networks, we designed a greedy algorithm with theoretical guaranteed-performance. For MapReduce-like systems, we designed algorithms for we study offline scheduling of minimizing makespan and minimizing total completion time, respectively. For client/serve system, we propose a new model for the client assignment problem and design an algorithm based on Semidefinite programming. For wireless data broadcasting, we designed an approximation algorithm to maximize the number of downloads given a deadline. For rumor blocking in social networks, we proposed two new models and designed approximation algorithms for minimizing the number of protectors in these two models. We organized a research seminar once a week. We had also continued to provide a course "cs7301 Advanced Algorithm Design in Wireless Sensor Networks" and a course "cs7301 Advance Algorithm Design in Social Networks" in Fall 2013 and Spring 2014. Meanwhile, we had discussion with our collaborators. There are ten journal papers published and seven conference papers published/accepted. Results in above publications are formed main parts of five Ph.D. dissertations of Yuqing Zhu, Lidan Fan, Lidong Wu, Yuanjun Bi and Kai Xing who received their Ph.D. degree in 2014. Our research has generated several novel algorithms, some of which improve on the performance and accuracy of previous algorithms, theoretically prove and showcase the optimal placement of data replica to enhance the security in distributed database, and provide an opportunity for the training of several graduate students. Most of our results have been published in various prestigious journals and conferences such as IEEE Transactions on Mobil Computing, the International Conference on Data Mining (ICDM), IEEE International Conference on Computer Communications (INFOCOM), etc.
