Availability Analysis of a Drone System with Proactive Offloading for Software Life-extension

Abstract

Real-time image processing on a drone to recognize the real-world environment has become popular recently in many applications. However, continuous image processing on a drone may entail the degradation of performance and reliability over the long-time operation, also known as software aging. Since the degradation due to software aging progresses with the amount of the workload to process, offloading the image processing tasks to other computers can mitigate the progression of the software aging. In this paper, we propose a new software life-extension method to counteract software aging on a drone image processing system by means of proactive task offloading. To evaluate the effectiveness of the proposed method, we develop continuous-time Markov chains (CTMCs) to analyze the stochastic behaviors of the system. Through numerical experiments, we show that proactive offloading improves the steady-state availability, the mean time to down (MTTD), and the average throughput by 1.85%, 1.57x, 1.48x, respectively. We also show that the combination of offloading and software rejuvenating further improves the steady-state availability and the average throughput.