Controlled mobility of unmanned aircraft chains to optimize network capacity in realistic communication environments

This dissertation presents a decentralized gradient-based mobility control algorithm for the formation and maintenance of an optimal end-to-end communication chain using a team of unmanned aircraft acting as communication relays. With the use of unmanned aircraft (UA) as communication relays, a common mode of operation is to form a communication relay chain between a lead exploring node (which may be ground based or another UA) and a control station. In this type of operation the lead node is typically deployed to explore (sense) a remote region of interest that is beyond direct radio frequency (RF) communication range, or out of line-of-sight, to the control station. To provide non-line-of-sight service, and extend the communication range of the lead node, unmanned aircraft acting as communication relays are deployed in a convoy fashion behind the lead vehicle to form a cascaded relay chain.;The focus of this work is the use of the mobility of a fixed number of relay aircraft to maximize the capacity of a directed communication chain from a source node to a destination node. Local objective functions are presented that use the signal-to-noise-and-interference ratio (SNIR) of neighbor communication links as inputs to maximize the end-to-end capacity of packet-based and repeater-type network chains. An adaptive gradient-based SNIR controller using the local objective function can show significant improvement in the capacity of the communication chain that is not possible with range-based controllers, or static deployment strategies, in RF environments containing unknown localized noise sources and terrain effects. Since the SNIR field is unknown, an online estimate of the SNIR field gradient is formed using methods of Stochastic Approximation from the orbital motion of the aircraft tracking a control point.;Flight demonstrations using the Networked Unmanned Aircraft System Command, Control and Communications testbed were conducted to validate the controller presented herein. Results from flight experiments show that mobility of unmanned aircraft, following locally estimated SNIR gradients, can be used to form a locally optimized communication chain by driving the aircraft to locations that improve the end-to-end capacity of the chain over that of a range-based controller or a static deployment algorithm.