Online Hybrid RF Propagation Model for Communication-Aware sUAS Relay Applications

摘要

To develop communication-aware robotics agents need a way of determining the RF propagation field online and without external assistance. An online architecture for a Hybrid RF Propagation model is created to provide the path loss necessary to become communication-aware. The Hybrid RF Propagation model is composed of an Irregular Terrain Model (ITM) for prediction and a Gaussian process to correct the prediction error from measured. During flight tests over the summer of 2017, the two components of the hybrid model are evaluated. The implementation of lTM called SPLAT! performed well at predicting the path loss In the field with 11 mean error of 0.23 dB. The Gaussian process did an excellent Job in learning the RSS flield, slow processing hinders use onboard the aircraft with current technology. However, initial analysis suggests parallelizing components of the Gaussian process is an effective strategy to reduce computational time. Simulations are also conducted evaluating the use of the hybrid model for sUAS relay applications between two nodes looking to maximize throughput. The results of the simulations show that the hybrid model consistently achieves slightly better performance than the midpoint between the two nodes, hut is far off from the best possible case.