Cloud computing is one of the most used distributed systems for data processing and data storage. Due to the continuous increase in the size of the data processed by cloud computing, scheduling multiple tasks to maintain efficiency while reducing idle becomes more and more challenging. Efficient cloud-based scheduling is also highly sought by modern transportation systems to improve their security. In this paper, we propose a hybrid algorithm that leverages genetic algorithms and neural networks to improve scheduling. Our method classifies tasks with the Neural Network Task Classification (N2TC) and sends the selected tasks to the Genetic Algorithm Task Assignment (GATA) to allocate resources. It is fairness aware to prevent starvation and considers the execution time, response time, cost, and system efficiency. Evaluations show that our approach outperforms the state-of-the-art method by 3.2% at execution time, 13.3% in costs, and 12.1% at response time.

翻译：云计算是最广泛使用的分布式数据处理与存储系统之一。随着云计算处理的数据规模持续增长，如何在保持效率的同时减少空闲资源，使多任务调度变得更加具有挑战性。现代交通系统为提升安全性，对高效的云调度方案也提出了迫切需求。本文提出一种融合遗传算法与神经网络的混合调度方法。该方法通过神经网络任务分类（N2TC）对任务进行分类，并将筛选后的任务提交至遗传算法任务分配（GATA）模块进行资源分配。算法具备公平性感知能力以避免任务饥饿，同时综合考虑执行时间、响应时间、成本及系统效率。评估结果表明，本方法在执行时间、成本和响应时间上较现有最优方法分别提升了3.2%、13.3%和12.1%。