Maximizing Sampling Frequency for UAV Clusters in Spectrum Reconnaissance Applications

Unmanned Aerial Vehicle (UAV) clusters are widely used in spectrum reconnaissance, where spectrum samples are transmitted back to the sink through flying ad hoc network (FANET). High sampling frequencies are required, but result in a large amount of data not being transmitted over the wireless network, affecting throughput and reliability of the network. The problem boils down to what a good sampling frequency is and how to find it. There is an unknown dynamic sampling frequency threshold, beyond which high packet loss occurs. Prior work focused on achieving higher frequencies through FANET optimization but neglected methods to find this threshold and achieve a maximum value. We propose an adaptive protocol based on binary search. With limited bandwidth and consistent sampling frequency, the protocol dynamically adjusts sampling frequencies in a phased manner at the application layer to maximize the sampling frequency that the network can handle. In addition to spectrum reconnaissance, this protocol can be extended to other applications with consistent sampling frequencies. We have constructed an application based on the EXata simulation platform to analyze the performance of the protocol in delay, stability, reliability, and search times. Simulation results show the protocol improves the efficiency and survivability of spectrum reconnaissance.