Leveraging 5G RedCap and Spiking Neural Networks for Energy Efficiency in Edge Devices

This work presents an energy efficient implementation for UAV-based systems over 5G networks with on-boarded accelerated processing capabilities and provides a preliminary evaluation analysis of the integrated solution. A two-fold comparative study focused on connectivity and edge processing for UAVs, realizes two discrete deployment scenarios, where standard 5G configuration with Artificial Neural Networks processing is evaluated against 5G RedCap connectivity paired with Spiking Neural Networks. Both proposed alternative energy efficient solutions, are designed to offer significant energy saving, and this paper examines if they are fit candidates for energy stringent environments, i.e., UAVs, and also quantify the impact on the overall energy consumption of the system. The integrated solution with 5G RedCap/SNN realizes energy-use reductions approaching 60%, which translated to approximately 35% of increased flight time. The experimental evaluations were performed in a real-world deployment using a 5G equipped UAV with edge processing capabilities based on NVIDIA’s Jetson Orin.