Providing network communication in large, complex environments is an important task with applications to maintaining connectivity with mobile users, environmental monitoring, emergency response, search and rescue, etc. Traditional approaches accomplish this task by deploying a static wireless network over the entire environment. However, this solution becomes cost ineffective when the area to be covered is large.
Recent advances in robotics technology and research have made it possible to build low-cost, robust mobile robots that can autonomously navigate in complex environments. Thanks to these advancements, it is now feasible to use robots as mobile network nodes in place of large static network deployments. However, in order to achieve good performance with a small number of robots, it is crucial to design efficient algorithms for planning the robots' paths. In this dissertation, we study the use of mobile robots in two specific networking applications.
In the first application, we use mobile robots to provide communication between end-points that require a persistent connection in a large, complex environment. For instance, imagine that a mobile user in a large environment requests connectivity to a static base station. Since the service area of wireless routers is limited by their initial deployment locations, a static wireless network deployment requires many routers to fully cover the entire region. Alternatively, this drawback can be overcome by using a small number of robots as intermediate mobile routers between the user and the base station which adaptively maintain connectivity according to the user's movement.
In the second application, we seek the use of mobile robots in delay-tolerant networks where a small delay in data transfer is acceptable. One such application is environmental monitoring where scientists collect statistical data such as soil temperature. The most crucial problem in this application is to gather the data from sensors which may be sparsely deployed over a large environment. We can avoid the inefficient use of intermediate relay nodes for data transfer by using mobile robots to autonomously collect the data from sensors. Since a small delay is tolerable, using a few robotic data carriers becomes an appealing solution.
Our contributions in this dissertation are two-fold: on the theoretical front, we present path-planning algorithms with provable performance guarantees. First, we study the problem of maintaining the connectivity of a mobile end-point to a static end-point by using the minimum number of mobile routers. Second, we present solutions for creating a communication bridge between two static end-points by minimizing the number of robots and their movements. Third, we study the problem of finding robot paths so that robots collect data from sensors as quickly as possible. Lastly, we present strategies for robots which act as mobile sensors to efficiently monitor an environment. On the systems front, we implement our algorithms using mobile robots and demonstrate their practical feasibility through extensive experiments.