Premium
An intelligent efficient scheduling algorithm for big data in communication systems
Author(s) -
Bu Fanyu
Publication year - 2017
Publication title -
international journal of communication systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.344
H-Index - 49
eISSN - 1099-1131
pISSN - 1074-5351
DOI - 10.1002/dac.3465
Subject(s) - computer science , reinforcement learning , big data , energy consumption , distributed computing , scheduling (production processes) , frequency scaling , mobile device , efficient energy use , real time computing , artificial intelligence , data mining , operating system , ecology , operations management , economics , biology , electrical engineering , engineering
Summary Recently, a large number of mobile computing devices with embedded systems have been widely employed for big data analysis in communication systems. However, mobile computing devices usually have a limited energy supply. Consequently, green and low‐energy computing has become an important research topic for big data analysis in communication systems when using energy‐limited mobile computing devices. In this paper, we propose a reinforcement learning‐based intelligent scheduling algorithm for big data analysis by increasing the utilization and reducing the energy consumption of the processors. Specially, we design a reinforcement learning model, as an important big data intelligent technique, to select an appropriate dynamic voltage and frequency scaling technique for configuring the voltage and frequency according to the current system state, which can improve the utilization and optimize the energy consumption effectively. Furthermore, we implement a learning algorithm to train the parameters of the reinforcement learning model. Our proposed scheduling approach is able to improve the resource utilization and save the energy for big data analysis in communication systems when performing tasks on mobile computing devices with embedded systems. Simulation results demonstrate that the proposed method can save 2% to 4% energy than the compared algorithm.