Wireless sensor networks (WSN’s) provide unprecedented spatial and temporal sensory resolution. The ubiquity of wireless sensor networks is made possible by their small size, Figure 1. These devices have remarkably low power consumption and once powered up, can operate without service for months, or years.
CRIM WSN’s research looks at large area applications requiring hundreds or thousands of nodes, i.e., for monitoring for wildfires, search and rescue tasks, and battlefield information gathering. For these situations it is not practical to optimally hand place each node, therefore an air-drop will be used to distribute the nodes. As the nodes become active, they will form an ad-hoc networks. Due to the random nature of their deployment several network clusters will be formed, Figure 2. Each node is 
specified by a letter or number with its communication range denoted by the colored circle surrounding it. Like colors represent node clusters that can communicate with one another. As more nodes are dropped, these networks will become connected but there is a point of diminishing return where the addition of numerous nodes will be needed where a few intelligently placed nodes would have been adequate to connect the disjoint networks.
Determining how to connect these networks is an interesting problem as there is little to no information regarding the location of the nodes within the deployment area. This research uses an autonomous robot, equipped with a radio, such that it becomes a mobile sensor node, Figure 3. By monitoring the communication signal strength between itself and any nodes in its broadcast range, the robot “ski’s” through a network cluster in search of a connection point and connects to other clusters. The robot adjusts its performance to deal with obstacles in its communication range. Upon completion of the repair task, the robot has improved network efficiency and coverage, and has reduced the number of required nodes by a factor of 10.
