Optimization of periodic synchronization of UAV's clock by differential phase method

The article considers the problem of designing a synchronized aerial wireless sensor network of unmanned aerial vehicles (UAVs) with centralized control by a master unit. A 10-ns synchronization precision must be ensured for all units within the aerial network in the time intervals of at least 100 s. To achieve it, the master unit periodically generates a synchronizing signal with two coherent sine tones of different frequencies that helps the slave UAVs to adjust their clocks. High-stability oven-controlled crystal quartz oscillators (OCXO) are used as onboard clocks for the UAVs.
The study aims to assess the optimal frequency separation of the coherent synchronizing tones that provides the best possible noise immunity of the measured data with a reliable ambiguity resolution of the carrier phase. The problem is solved using a computer simulation of periodic synchronization of the slave UAVs by differential phase measurements associated with the reference time scale of the master UAV in order to suppress possible random clock offsets.
According to the simulation results, aside from the positioning errors of the UAVs, the systematic Doppler phase shift of the synchronizing signal in the propagation channel is the main obstacle to differential phase synchronization. Depending on the efficiency of the Doppler phase shift compensation, the optimum frequency separation of the synchronizing tones ranges from 10 to 1500 MHz with the correspondent synchronization precision achieved from 0.25 to 2.65 ns at the observation time of up to 100 s. It is shown that the effective compensation for the Doppler shift requires periodic channel estimation for at least every 10 ms. For most practical applications, however, adjusting the slave clock every 5 s using two coherent synchronizing sine tones separated by 400–500 MHz results in a satisfactory quality of synchronization.