Cooperative image analysis in visual sensor networks based on Nash bargaining solution

One of the most dynamically evolving areas in Telecommunications is the Internet of Things. This concept describes "a world-wide network of uniquely addressable and interconnected objects, based on standard communication" (ETP-EPOSS). The Internet of Things can find application practically everywhere starting from a wearable device monitoring steps to a space shuttle coffee machine. This thesis research considers visual sensor networks (VSN), one of applications of the Internet of Things. VSNs enable a smart surveillance system that can improve areas of early emergency alarming, safety and health-state monitoring. Taking into consideration battery-limited lifetime of sensor network the topic of energy saving is of actual importance. This thesis work defines an outline, where a visual sensor network is densely populated with camera nodes, in case of surveillance of public areas application. In this reference scenario some of the cameras have an overlapped field of view. Furthermore each camera performs visual analysis task: acquire image, extract features, transmit compressed features over the network. Therefore camera nodes acquire, process and spend limited energy resources on the same (or partially the same) image. This work proposes cooperative framework in which camera nodes with (partially) overlapped field of view collaborate to reduce the overall energy consumption. Camera nodes collaborate upon a portion of image they have to process, playing a bargaining game. The optimal solution to this game is offered by the Nash bargaining solution (NBS). NBS implementation allows to obtain "efficient" image processing sharing policies.