A number of algorithms[19,21,20,18] - mainly simulations - have been used to evaluate deployment of sensor networks using mobile sensor nodes in order to maximise network coverage.  models a sensor network where all nodes have locomotion capability and are equipped with sensors that can determine the range and bearing of both nearby nodes and obstacles. The nodes are modelled as particles that interpret their distance to other nodes and obstacles as a virtual force which repels them from one another. This allows for the creation of an evenly distributed network while increasing the network coverage as much as possible. To ensure the repelling is not done to the extent where nodes would fail to communicate with one another, a conceptual firm friction force is assumed to be acting on the nodes which avoids them from moving too far apart.  Considers use of static and mobile nodes in the same network where static sensor nodes detect coverage holes and compete for acquiring a mobile node to cover that area. The mobile nodes will analyse various requests for coverage and move to the areas which maximises the network coverage while taking into account movement and communication overheads.
 Utilises sensor nodes' mobility to relocate nodes in the network in order to replace failed sensors. This avoids creation of holes in the network which could potentially result in disconnected islands of sensor nodes. The relocation of nodes are done in a cascading manner which has two advantages; first the hole in the network gets covered quicker, second, the energy consumption of the relocation task gets spread out between the nodes participating in this task whereas if a single node was to perform this task it might have needed to travel a long distance and therefore consumed too much power by the time it reached the destination which would result in this relocated node failing(due to running out of power) soon after taking its new place.
 Looks at simulation of a sensor network with a small number of mobile robots relative to the static nodes. Mobile robots have the responsibility of replacing failed nodes and depending on the algorithm used for replacement of nodes the mobile robot might also be in charge of making decisions as to which mobile robot is to replace a failed sensor. Three algorithms are examined in these simulations; i) A central algorithm in which one robot does all the management task and assigns which robot should replace the failed node. ii) A distributed algorithm in which robots are spread out during the initialisation stage and the static sensor nodes choose one of the mobile robots that is closest to them as their maintenance robot which will be responsible if they fail. iii) Another distributed algorithm similar to the previous one, however in this one the robots broadcast their location as they are moving to replace a failed node. This allows static sensor nodes to select another robot that is closer to them to be their maintenance robot at any given moment in time. Each of the mentioned algorithms have their own advantages in terms of minimising communication and locomotion overhead as well as maximising scalability.
Some work has been done on using static sensor networks in conjunction with mobile robots in order to navigate a mobile robot to a desired location. Other work on localisation and navigation includes one where RFID tags were used to more accurately localise a mobile robot. In the field of robotics navigation of robots have been improved in indoor narrow spaces such as hallways which in domain of sensor networks translates to a decrease in power consumption and faster relocation of mobile robots.
 Explores use of static nodes together with a single mobile robot for a home security system in which various emergency events are detected by the sensor network. The robot is informed which takes pictures of the location in which the event was detected and a home server sends these images along with the event details to the user's hand-held device.
 Presents the design of a mobile sensor node. This platform is approximately 0.000047 cube meters and it costs approximately $150. Even though this platform is relatively cheap and very small which makes it appealing for use in sensor networks, it can not be used in rough environments because its mobility feature has been compromised to favour lower power consumption and its small size.
Jenson Taylor 2008-01-25