In mobile cloud computing, the computing resources of mobile devices can be integrated to execute complicated applications, in order to tackle the problem of insufficient resources of mobile devices. Such applications are, in general, characterized as workflows. Scheduling workflow tasks on a mobile cloud system consisting of heterogeneous mobile devices is a NP-hard problem. In this paper, intelligent algorithms, e.g., particle swarm optimization (PSO) and simulated annealing (SA), are widely used to solve this problem. However, both PSO and SA suffer from the limitation of easily being trapped into local optima. Since these methods rely on their evolutionary mechanisms to explore new solutions in solution space, the search procedure converges once getting stuck in local optima. To address this limitation, in this paper, we propose two effective metaheuristic algorithms that incorporate the iterated local search (ILS) strategy into PSO and SA algorithms, respectively. In case that the intelligent algorithm converges to a local optimum, the proposed algorithms use a perturbation operator to explore new solutions and use the newly explored solutions to start a new round of evolution in the solution space. This procedure is iterated until no better solutions can be explored. Experimental results show that by incorporating the ILS strategy, our proposed algorithms outperform PSO and SA in reducing workflow makespans. In addition, the perturbation operator is beneficial for improving the effectiveness of scheduling algorithms in exploring high-quality scheduling solutions.

中文翻译：

---

在移动云上调度工作流任务的有效算法

在移动云计算中，可以整合移动设备的计算资源来执行复杂的应用程序，以解决移动设备资源不足的问题。此类应用程序通常以工作流为特征。在由异构移动设备组成的移动云系统上调度工作流任务是一个 NP-hard 问题。在本文中，智能算法，例如粒子群优化（PSO）和模拟退火（SA），被广泛用于解决这个问题。然而，PSO 和 SA 都受到容易陷入局部最优的限制。由于这些方法依赖于它们的进化机制来探索解决方案空间中的新解决方案，因此一旦陷入局部最优，搜索过程就会收敛。为了解决这个限制，在本文中，我们提出了两种有效的元启发式算法，它们分别将迭代局部搜索 (ILS) 策略合并到 PSO 和 SA 算法中。在智能算法收敛到局部最优的情况下，所提出的算法使用扰动算子来探索新的解决方案，并使用新探索的解决方案在解决方案空间中开始新一轮的演化。重复此过程，直到无法探索更好的解决方案。实验结果表明，通过结合 ILS 策略，我们提出的算法在减少工作流制造时间方面优于 PSO 和 SA。此外，微扰算子有利于提高调度算法探索高质量调度方案的有效性。