Cloud Computing for Big Data Processing

The topics of this special issue are mainly devoted to the most recent research, development and applications in the field of cloud computing, big data processing in the various fields, e.g. Web blog, search engine, image processing, and industry. Among all of the submitted manuscripts, 15 papers were selected and included in this special issue. Creative thoughts and interesting inspirations are presented, discussed, and disseminated in this issue.

First, there are three papers on the cloud architecture and optimization for big data processing. J. Peng, et al. discussed the problems of load balance for massive data processing on cloud considering the data processing efficiency and nodes’ capability. A load balance approach for massive data, based on the consistent hashing method, was presented to improve the data processing equalization and high processing efficiency. Large-scale sensor networks and internet of things are one of the most important application fields in big data. Y. Sun, et al. proposed a cloud computing system to support PaaS in energy power applications—PROXZONE. PROXZONE provides the cloud services uniform monitoring and authorization, and cloud message service, which can support good performance when dealing with big data. Meanwhile, S. Xu, et al. discussed an improved cooperative dynamic cluster model in multitasks mobile network, and proposed a cooperative message forwarding mechanism to improve the network performance.

Two papers focus on big data storage and query processing in wireless sensor networks. J. Guo, et al. considered distributed storage issues for big data from the IOT (Internet of Things). They addressed the k-storage-node problem and converted it to a k-median problem in a non-metric space. Then, distributed storage solutions including random strategy based, reverse greedy based, and small world model based quantum genetic algorithm were proposed. Y. Mao, et al. discussed the uncertain data query processing for IOT in the dam safety monitoring. Based on the x-tuple Rule of uncertain data, using intra-cluster and inter-cluster two phases query processing, a distributed Two-Phase PT-Topk Query Processing approximation algorithm (TPQP) was proposed to handle the large-scale uncertain data processing. Localization is one of main issues in the wireless sensor networks.

Web applications are the important source of big data. Five papers addressing Web data analysis were selected for this special issue. In the Web search engine application, to eliminate the deviation of statistics about the user’s intent, H. Zhou, et al. proposed a new method that can identify Web pages with the same contents, (but with different URLs) by calculating similarity among contents of traversed pages and non-traversed pages. Each component search engine weight can be obtained dynamically when a search service is provided. Thus, the QoS of search services can be improved facing the urgent demand of Web data applications. W. Wu, et al. addressed the analysis and mining of big Web data based on the Cloud. They explored Web mining algorithms and Cloud Computing technologies to analysis and processing of mass web data based on Cloud Computing platform. H. Zhang, et al. discussed how to efficiently exact the interest association rules from big data mining. Considering the users’ interests and the support of all association rules, they proposed the multiple minimum supports and weights assignment method to exact the appropriate associate rules. The experimental results evaluated the efficiency of the proposed methods. Y. Wang, et al. proposed the Chinese hot words detection method based on Chinese semantic clustering. To analyze the sparse and content fragmentation features of Chinese WeChat and blog data, they applied the repeated string computation, context adjacency analysis and linguistic rule filtering strategies to abstract and express the complete semantics. Then candidate topic generation rules were proposed via feature clustering and heating sorting approaches to reduce the dimension of WeChat and blog data. G. Xu, et al. considered the semantic-based Web Services matching issues in Web data processing. To improve the accuracy of Web Services matching, the data provenance is regarded as a constraint attribute of Web Services matching. An SP-tree-based Web Services matching algorithm was proposed and a prototype system in the water resources applications was implemented.

As to the cloud applications in the field of industry, four papers were selected and included to this special issue. To effectively predict and control the quality of products, J. Qiu, et al. presented a data processing framework based on Hadoop and adopted LS-ANN (Least Square Artificial Neural Network) and a time series analytical method to process and analyze the big data industry. Y. Miao, et al. discussed the PID controller in the control engineering field. To avoid the shortcomings of traditional data reconciliation methods depending on the mathematical model or particle swarm optimization (PSO) algorithm, one improved PSO algorithm based on simulated annealing was proposed to avoid the local best and improve the precision when dealing with a large-scale data-set. Y. Miao, et al. discussed the extreme machine learning (EML) for prediction. Due to over fitting of the existing EML approaches based on Empirical Risk Minimization, it can result in poor performance when there are many outliers in data integration. To solve these problems, one integrated method combining PCA (Principal Component Analysis) and ELM was proposed to simplify the structure of the Feedforward Neural Network, and improve the prediction precision. X. You, et al. focused on the parameters re-configuration in the cognitive radio system. They adopted an optimal solution based on the multi-objective artificial bee colony and fuzzy reasoning algorithm. The fuzzy reasoning method takes the job to select the optimal solution from the Pareto solution set to meet the users’ requirements. The solution to the multi-object optimal algorithm was processed in parallel with the cluster. The experimental results indicate the efficiency of the proposed optimal solution.

Finally, in this special issue, the last two papers are about large-scale image processing. To reduce the high dimensions of hyperspectral data for image classification, an application of particle swarm optimization (PSO) based on image information content and between-class separation criteria was proposed to band selection of a hyperspectral image. M. Xu, et al. proposed an algorithm for tracer particles detection and image data processing of complex water surfaces based on the analysis of the tracer particles and the reflection noise over a water surface. Then, a combination of Top-Hat transform and adaptive threshold segmentation is utilized to detect floating small objects, and fine matching is carried based on the similarity among the particles and motion distance without error vectors. J. Zhao, et al. discussed the problem of self-learning intelligent single particle optimization. They presented a self-learning intelligent single particle optimizer based on the adaptive image de-noising in a shearlet domain. The proposed approach can adaptively determine the optimal thresholds of different scales and directions and achieve image content-based adaptive de-noising.

Xiaofang Li
School of Computer Information and Engineering, Changzhou Institute of Technology, Changzhou, China
Yanbin Zhuang
School of Computer Information and Engineering, Changzhou Institute of Technology, Changzhou, China
Simon X. Yang
Advanced Robotics and Intelligent Systems Laboratory, School of Engineering, University of Guelph, Guelph, Canada
[email protected]