The new-generation of Internet of Things (NG-IoT) brings a wide range of challenging problems. At the same time, cloud computing technology is an important foundation for the development of the IoT. In this article, we focus on the task scheduling problem in IoT systems in cloud computing environment. Our goal is to minimize the task runtime. It is well known that the problem of the task scheduling has been a challenging problem. In the last decade, despite being theoretically hard problem, researchers design lots of state-of-the-art algorithms for solving this problem. In our work, we propose a novel efficient reinforcement learning (RL) algorithm to solve the task scheduling problem in IoT systems (EATS), which combines combinatorial optimization to make our proposed algorithm have stable lower bounds. We process a batch of tasks at a time, make decisions on task selection through reinforcement learning, and solve them further through combinatorial optimization methods. The results of the experiments show that our proposed algorithm has outstanding performance in different environments.
The development of Internet of Things (IoT) technology depends on technologies such as high-efficiency storage and high computing power. Mobile cloud computing (MCC) technology will be an important foundation for the development of IoT. The efficient scheduling of tasks in IoT devices in MCC environment is challenging. The requirements for task scheduling in MCC are becoming more and more complex. As the core problem in MCC, task scheduling aims to allocate tasks reasonably, achieve optimal scheduling strategies, and complete tasks effectively. In this paper, efficient delay-aware task scheduling algorithm (EDTSA) is proposed, with the optimization goal of minimizing task running time. The matching of tasks and virtual machines is modeled as a bipartite graph. The problem is divided into multiple subproblems to solve the optimal solution separately. The combined solution is used as the initial solution of the local search algorithm. The efficiency of the local search depends on the quality and nature of the initial solution. We can generate multiple initial solutions according to different division criteria. The initial solution is the combination of the optimal solutions of the subproblems, so the quality of the initial solution has been greatly improved and generating multiple initial solutions according to the division can reduce the probability of falling into the local optimal solution. This algorithm also divides the neighborhood to reduce unnecessary searches. Finally, we verify the efficiency and practicability of the algorithm through experiments.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.