Computation offloading for object-oriented applications in a UAV-based edge-cloud environment

The high mobility and maneuverability of unmanned aerial vehicles (UAVs) enable them to act as temporary base stations (BSs) in extreme environments, expanding the computing capacities of terminals for intelligent applications, most of which are object-oriented ones. Computation offloading in a UAV-based edge-cloud environment is an excellent way to improve the performance of these object-oriented intelligent applications. In contrast, the computation-intensive tasks are offloaded to the cloud, and the data-intensive ones are offloaded to the edge. Though computation offloading over the cloud, edge, and terminals has been broadly studied, existing researches primarily establish scheduling algorithms on program high-level abstraction without consideration of challenges from program structures. We focus on task scheduling for offloading object-oriented applications while considering the 'encapsulation' characteristic. We proposed a time-driven offloading strategy based on a particle swarm optimization algorithm employing the genetic algorithm operators with floating encoding (PGFE). This strategy introduces a genetic algorithm's randomly two-point crossover and mutation operator to avoid converging on local optima effectively. The simulation results show that our strategy can reduce the average execution time of object-oriented applications by 11.78-48.02%, compared with other classic algorithms in a UAV-based edge-cloud environment.