Joint Optimization of Trajectory-Resource Allocation and Deep Task Partial Offloading for MEC-Enabled Multi-UAV.

Area of Science:

• Computer Science
• Electrical Engineering
• Robotics

Background:

• Mobile edge computing (MEC) with multiple unmanned aerial vehicles (UAVs) offers computation services to ground terminals (GTs).
• Deep learning tasks in challenging environments require efficient offloading strategies.
• Existing systems face degraded Quality of Service (QoS) due to blocked UAV-GT links, limited UAV capacity, and energy constraints.

Purpose of the Study:

• To address degraded QoS in MEC-enabled multi-UAV systems by jointly optimizing UAV trajectories, computing resource allocation, and partial program offloading (PPO) for deep learning tasks.
• To minimize task latency for ground terminals (GTs) in complex environments.

Main Methods:

• Formulated a joint optimization problem for UAV trajectories, resource allocation, and partial task offloading.
• Employed Successive Convex Approximation (SCA) and Block Coordinate Descent (BCD) to solve the non-convex problem.
• Evaluated the proposed scheme against benchmark solutions through numerical simulations.

Main Results:

• The proposed joint optimization scheme significantly reduces task latency for GTs.
• The method effectively manages UAV trajectories, resource allocation, and partial offloading decisions.
• Achieved superior performance compared to existing benchmark solutions in MEC-enabled multi-UAV systems.

Conclusions:

• Joint optimization of UAV trajectories, resource allocation, and partial offloading is crucial for enhancing QoS in MEC-enabled multi-UAV systems.
• The SCA and BCD approach provides an effective solution for complex optimization problems in this domain.
• This research offers a promising direction for improving deep learning task processing for GTs in challenging environments.