Section: New Results

Controlled mobility based services

Participants : Emilio Compagnone, Valeria Loscri, Karen Miranda, Nathalie Mitton, Tahiry Razafindralambo, Dimitrios Zormpas, Jean Razafimandimby Anjalalaina.

Sensors have more and more functionality in terms of capture techniques, communication capabilities, processing capabilities and energy harvesting. Another interesting feature available on sensors is mobility. The FUN research group tries to exploit the controlled mobility of sensors to solve some known issues in wireless sensors networks regarding deployment or routing but also raises some new challenges regarding coverage optimization and energy harvesting.

Coverage

Wireless sensors are used to gather information from a field of interest. In order to capture all the events in this field, the sensors must be properly placed. When the sensors have motion capabilities such as robots, the deployment can be optimized. The use of controlled mobility raises some new challenges and opportunities in the field of wireless sensor networks. Milan Erdelj and Karen Miranda in [33] presents the advances in context. They provide a detailed literature review regarding the techniques behind controlled mobility in order to deploy or redeploy sensors. When the wireless sensors are mobile, it is possible to optimize the capture of information regarding their time and space evolution. This allows the sensors to focus on different zones of interest depending on the evolution of the observed events. Valeria Loscri, Enrico Natalizio and Nathalie Mitton present a performance evaluation of different algorithms for zone of interest coverage in [18] . Their work particularly focuses on providing a set of distributed version of a combined particle swarm optimization and virtual forces algorithm. The proposed algorithms and their evaluation show an high reactivity to changing events and targets. Energy is an important constraint in wireless sensor networks and message exchange is a functionality that drains huge amount of energy. Dimitrios Zorbas and Christos Douligeris in [30] present a low-overhead localized algorithm for the target coverage problem in wireless sensor networks. To tackle this problem they propose two variations of a localized algorithm with low communication complexity in term of message exchange. The results show a great improvement in terms of communication cost while achieving an adequate network lifetime.

Connectivity and performance

Information gathered by sensors are to be processed in a remote location. The transportation from the point where the raw data is generated (the sensor) and the data processing unit (sink or other infrastructure) relies on routing techniques. Routing is a fundamental functionality of a wireless sensors network. Nicolas Gouvy, Nathalie Mitton and David Simplot-Ryl in their book chapter [34] provide a review of the routing techniques described in the literature. They highlight the challenges, main issues and future work direction in this domain and provide some important assumption and characteristics that should be kept in mind when designing routing protocols for wireless sensor networks. When route between a source and the destination of data does not exist or cannot be established, using a mobile router is a possible solution. Christos Katsikiotis, Dimitrios Zorbas and Periklis Chatzimisios in [15] propose an algorithm that restores connectivity by the use of mobile wireless router after a routing failure. They provide a fast mechanism to heal the network and restore connectivity between the network partitions. In their solution, a mobile wireless router finds the end points that should be re-connect and place itself in the correct position to restore the connectivity. Their solution shows a fast restoration process based on the implementation done on a real robotic platform.

Energy suppliance

Energy is an important constraint in static wireless sensor networks and even more important when sensors are mobile. However, when sensors have motion capabilities, they can use this ability to move toward a recharging point in order to increase the network operation. Dimitrios Zorbas and Tahiry Razafindralambo in [31] use the motion capability of sensors to provide an algorithm that allow the sensor to go to a recharging point while minimize the impact of their movement on the network operation such as portioning or data gathering. They provide theoretical bounds on the realisation of such operation and evaluate the average behaviour of their algorithm based on extensive simulations. Both results show a big improvement in terms of network lifetime extension compared to the case where no replacement is performed and to the case where rerouting is considered.

Video-based applications

Video Surveillance and Target Detection represent key components for many organizations in terms of safety and security protocols. The value of Video Surveillance has become more sophisticated and very accurate, by leveraging specific sensors able to detect motion, heat, etc. In [17] , Valeria Loscri, Michele Magno and Rosario Surace show how the nodes of a sensor network can learn which is their best position based on a certain number of WebCams that need to be "woken-up" when a suspicious event is detected. The main purpose is to reduce power consumption, especially in the case of Video Surveillance, when the most of the time the power is wasted by doing nothing. On the other hand, Target Detection, namely determining whether or not a target object exists in a video frame, has grown significantly with the recent advances in embedded computing and sensors which have opened the possibility to realize smaller and low-cost autonomous systems. In [16] , Valeria Loscri, Nathalie Mitton and Emilio Compagnone show the feasibility of low-cost embedded system for detection of objects based either on the shape or on the color.