Cooperative Planning for UAS-based Sensor Networks In Realistic Communication Environments

This dissertation develops a unified, tractable algorithm for distributed planning for multiple unmanned aircraft based sensor networks with performance bounds. An information-theoretic objective function is derived that incorporates sensing and communication to guide a cooperative team of unmanned aircraft. The communication is modelled with packet erasure channels for each link in a multi-hop mesh network. This objective is shown to be intractable and assumptions are made to find a tractable formulation. This tractable formulation is then distributed using the chain rule of mutual information and optimization algorithms are designed with performance bounds for different scenarios and compared against the fully centralized mutual information approach. This comparison is done analytically using submodularity under certain assumptions and Monte Carlo simulations to evaluate the algorithms when these assumptions break down. A novel unmanned aircraft system platform was developed to facilitate experiments with multiple unmanned aircraft utilizing sensing and multi-hop mesh networking. This system was used to perform localization experiments of radio frequency emitters based on the received signal strength measurement. The planning algorithms were demonstrated with multiple aircraft to show there validity and tractability on a real world scenario. These experiments were used to asses the algorithms performance showing the improvement in sensing is appreciable and specifically the benefits of utilizing multi-hop communication