Editorial for special issue on Big Data and knowledge extraction for Cyber-Physical Systems 2016

The cyber-physical system (CPS) has been coming into our view and will be applied in our daily life and business process management. The emerging CPS must be robust and responsive for its implementation in coordinated, distributed, and connected ways. It is expected that future CPS will far exceed today’s systems on a variety of characteristics, for example, capability, adaptability, resiliency, safety, security, and usability. Big Data are large, complex, or rapidly generated data sets that cannot be processed by traditional technologies. Every day, the entities comprising decision makers, managers, engineers, scientists, and citizens are faced with a multitude of constantly flowing data streams coming from different sources in different formats. Making sense of these volumes of Big Data requires cutting-edge tools that can analyze and extract useful knowledge from vast and diverse data streams. The wonderful living of humans and high efficiency of business rely mostly on how to use the Big Data intelligently and correctly and how to retrieve useful knowledge from the massive data—then, it would be possible to seamlessly integrate the virtual world and the physical world. How to integrate and analyze the data? How to retrieve knowledge from Big Data? How to share knowledge among smart things? How to ensure security and protect privacy? These are some of the questions in the long list of challenges that are needed to be addressed in future CPS.

This Special Issue solicits high-quality contributions with consolidated and thoroughly evaluated research in the area of Big Data and knowledge extraction for CPS that are worthy of archival publication in the journal. It is intended to (1) provide a summary of research that advances knowledge acquisition and utilization from Big Data for CPS and (2) serve as a comprehensive collection of the current state-of-the-art technologies within the context.

Due to the great effort made in publicity, we received 37 submissions from both academia and industry in the relevant fields. Following a strict review process, we accepted 22 papers for this Special Issue. Each of the papers was peer-reviewed by at least two experts in the field. In the following, we provide a brief introduction to each paper.

The paper “Parallel Irregular Fusion Estimation Based on Nonlinear Filter for Indoor RFID Tracking System,” authored by Xue-Bo Jin, Chao Dou, Ting-li Su, Xiao-fen Lian, and Yan shi, proposes a real-time indoor RFID tracking system which employs an irregular estimation strategy with parallel structure, where the dynamic model update and state fusion estimation are synchronously processed. Based on the extended Kalman filter (EKF) and unscented Kalman filter (UKF), two nonlinear estimation methods are designed. The tracking performance is evaluated, and the simulation results show that the UKF method achieves lower covariance in indoor RFID tracking, while the EKF one has less calculation time.

To efficiently track multi-level packages, the paper “Multi-level Package Identification Scheme Based on RFID Code and Related Package Message” by Yanghua Gao, Zhihua Zhang, Huanwen Wang, and Hailiang Lu proposes an RFID code scheme. With the investigation of the relation among multi-level packages, the subordinate relation of different level RFID codes can be established by recording and storing the inclusion relation between codes in database. Furthermore, to verify the feasibility of the proposed code scheme, it is applied to the fast-moving consumer goods management which involves three-level packages.

In the paper “On Energy-balanced Backpressure Routing Mechanisms for Stochastic Energy Harvesting Wireless Sensor Networks” by Zheng Liu, Xinyu Yang, Peng Zhao, and Wei Yu, the unpredictablility of the harvestable energy is considered, and the stochastic Lyapunov optimization framework is employed to jointly manage energy and make routing decisions. The purpose is to mitigate the energy imbalance problem. Two distributed online policies are developed, which are as follows: (1) Energy-balanced Backpressure Routing Algorithm (EBRA) for lossless networks and (2) Enhanced Energy-balanced Backpressure Routing Algorithm (EEBRA) for time-varying wireless networks with lossy links. Both of them are queuing stable and do not require the explicit knowledge of the statistics of energy harvesting.

In the paper “Efficient Sleep Scheduling Algorithm for Target Tracking in Double-storage Energy Harvesting Sensor Networks” by Hongbin Chen, Qian Zeng, and Feng Zhao, an efficient sleep scheduling algorithm is proposed to trade-off energy efficiency and tracking performance in energy-harvesting sensor networks. A double-storage energy-harvesting architecture is addressed as well to extend network lifetime. The simulation results indicate that the proposed algorithm can improve tracking performance and prolong network lifetime.

To reduce bandwidth contentions among different kinds of devices in CPS, the paper “Gateway Selection Game in Cyber-Physical Systems,” authored by Hao Wang, Jianzhong Li, and Hong Gao, formulates the gateway selection problem as a non-cooperative game. The authors investigate the actions of devices when their gateways change and the result of device competition. A gateway bandwidth allocation model is first developed, and a distributed algorithm is designed for clients in order to increase the total bandwidth of their own kind. The migration trends of clients are then investigated, and the authors provide some theorems regarding the conditions under which the clients stop migrating. Some examples of the gateway selection game with and without Nash equilibrium are illustrated.

The paper “Data Inconsistency Evaluation for Cyber-Physical System,” authored by Hao Wang, Jianzhong Li, and Hong Gao, investigates data inconsistency evaluation. This is the first study on inconsistency evaluation for CPS based on conditional functional. The authors first perform a thorough analysis on the complexity and inapproximability of the minimum culprit problem. It is shown that accurate evaluation is hard even for very simple cases. Therefore, a practical approximation algorithm is designed. For large dynamic data, the authors propose a compact structure based on B-tree to store independent residual subgraphs, so that inconsistency can be identified efficiently. Furthermore, additional extensions and optimizations are considered. The experiments are carried out toward both synthetic and real-life data sets, and the experiment results demonstrate the scalability of the proposed algorithm and the evaluation quality.

The paper “Cyber-Physical-Social-Thinking Modeling and Computing for Geological Information Service System” by Yueqin Zhu, Yongjie Tan, Ruixin Li, and Xiong Luo investigates a novel modeling and computing method for geological information service systems with the consideration of complex data processing for geological services under dynamic environment. The proposed techniques are implemented toward CPS and Internet of things. Two case studies are presented to show the efficiency of the proposed work.

In the paper “Cooperative Networking Towards Maritime Cyber Physical Systems” by Tingting Yang, Hailong Feng, Chengming Yang, Zhonghua Sun, Jiadong Yang, Fan Sun, Ruilong Deng, and Zhou Su, an innovative paradigm named Cooperative Cognitive Maritime Cyber Physical System is developed to achieve high-speed low-cost communication services. The authors present a bi-level game with two stages PUs-to-SUs and SUs-to-SUs to allocate resource. The Stackelberg game with priority is employed for PUs-to-SUs, and a symmetrical system model is considered for SUs-to-SUs. The simulation results demonstrate that the proposed strategy could effectively increase throughput as well as the payoffs of the system.

To enhance the estimation of member cardinality in cluster-based vehicular CPS, the paper “Efficient Evaluation of Member Cardinality in Cluster-based VCPS” by Jin Qian, Tao Jing, Yan Huo, Hui Li, and Zhen Li proposes a new member cardinality evaluation method based on the time slot occupation technique. The proposed method can improve the efficiency and accuracy of the estimation of member cardinality, which is an important parameter for cluster maintenance and update in dynamic vehicular CPS. The mathematical analysis and simulation results are presented to demonstrate the performance of the proposed method.

To improve the forecast accuracy of travel time in traffic management, the paper “A Kind of Urban Road Travel Time Forecasting Model with Loop Detectors,” authored by Guangyu Zhu, Li Wang, Peng Zhang, and Kang Song, develops a model for the massive data collected by loop detectors based on the analysis of the change-point and the autoregressive integrated moving average (ARIMA) model. A data preprocessing algorithm is first designed, and a travel time calculating model is constructed. Then, a change-point detection algorithm is proposed to classify the large number of travel time data into several patterns. Finally, a travel time forecast model is established based on the improved ARIMA model. The simulation results verify that the proposed work is practical and achieves high accuracy.

The paper “Hybrid Secure Beamforming and Vehicle Selection Using Hierarchical Agglomerative Clustering for C-RAN based V2I Communications in Vehicular Cyber-Physical Systems” by Dongyang Xu, Pinyi Ren, Qinghe Du, and Li Sun proposes a hydrid beamforming and vehicle selection framework for vehicle-to-infrastructure (V2I) communications to broadcast high-speed confidential messages. The authors consider a one-dimensional roadway with multiple lanes. Secure region and interference-selection region are, respectively, developed for each roadside unit. A secure beamformer for each vehicle located in the secure region is obtained to reduce interference and leakage of confidential information, based on which the vehicle selection problem is solved by employing a hierarchical agglomerative clustering method. The simulation results demonstrate the effectiveness of the proposed work on the aspect of secrecy sum rate.

The paper “Mobility-assisted Big Data Collecting in Wireless Sensor Networks” by Jinghua Zhu, Xuming Yin, Jingsi Bai, and Yake Wang studies how to efficiently collect big sensing data in wireless sensor networks. The authors propose a four-phase mobility-assisted data collection protocol, which involves network clustering, route planning, route combination, and data collecting. Two heuristic route planning algorithms are designed to build a set of trajectories considering the deadline constraint. The overall movement cost can be minimized. The authors also present numeric simulation results to show that the proposed approaches have better performance on the aspects of energy conservation, latency, and movement cost.

Another work addressing data collection in sensor networks is the paper “An Optimized Data Obtaining Strategy for Large-Scale Sensor Monitoring Networks” authored by Yan Wang, Junlu Wang, Fengtong Wang, Ling Wang, and Wei Wei. An optimized obtaining strategy is proposed in this paper for large-scale sensor monitoring networks, involving a large-scale monitoring network area clustering optimization strategy and a data acquiring strategy based on an adaptive frequency conversion. Furthermore, a linear regression model is developed which can modify the sampling frequency dynamically and provides a compensation mechanism for the sensor data model. The experimental results indicate that the proposed work can prolong network lifetime, reduce transmission cost and energy consumption, and improve network efficiency.

For target detection in wireless sensor networks, the paper “Extracting Target Detection Knowledge Based on Spatio-temporal Information in Wireless Sensor Networks” by Tian Wang, Zhen Peng, Cheng Wang, Yiqiao Cai, Yonghong Chen, Hui Tian, Junbin Liang, and Bineng Zhong adopts a practical signal model to describe the sensing process of sensors and a probabilistic decision model to make final decisions. Furthermore, a probabilistic detection algorithm is proposed which takes advantage of all sensors’ local measurement values. Spatiotemporal information can be utilized to make final decisions. The extensive simulation results show that the proposed method has high detection probabilities and low false alarm probabilities.

The paper “Proper Global Shared Preference Detection based on Golden Section and Genetic Algorithm for Affinity Propagation Clustering” by Rongfang Bie, Libin Jiao, Guangzhi Zhang, Rashid Mehmood, and Shenling Wang proposes a procedure “Golden Section and Genetic Algorithm for Affinity Propagation (GS/GAAP)” for global shared preference selection, which involves identifying a proper cluster number for clusters based on golden section and genetic algorithm. A global shared preference can be obtained based on the golden section value between the minimum and maximum similarities for affinity propagation, a well-known effective clustering algorithm. Then, a clustering result becomes robust with the global shared preference. One simulation data set and eight standard benchmark data sets are employed to verify the effectiveness of the proposed algorithm. The experiment results indicate that GS/GAAP outperforms the original affinity propagation clustering algorithm.

In the paper “Novel Individual Location Recommendation with Mobile Based on Augmented Reality” by Zhenghao Shi, Hao Wang, Wei Wei, Xia Zheng, Minghua Zhao, Jinwei Zhao, and Yinghui Wang, a mobile augmented reality-based system is proposed for individual location recommendation. The proposed system can locate users, detect markers, has 3D display, and provides location guidance by combing augmented reality with navigation technologies. In this system, the position and orientation information can be obtained to enhance the association to the real world. The experimental results show the effectiveness and efficiency of the proposed system in helping people locate their positions.

To protect privacy of smartphone users, in the paper “Providing Privacy Protection and Personalization Awareness for Android Devices” by Hongliang Liang, Dongyang Wu, Shirun Liu, Hao Dai, and Haifeng Liu, a new personalization aware privacy protection architecture is proposed. This architecture can support applications’ personalization functions in Android according to users’ personae and prevent users’ sensitive information to be accessed without authorization. Furthermore, a privacy protection service for legacy apps is designed, which can enforce different protection policies according to the risk level of the apps. A prototype system named P3Android is implemented. Experiments are carried out to show that P3Android is feasible and effective.

To handle concept drift for data streams, the paper “Online Ensemble Using Adaptive Windowing for Data Streams with Concept Drift” by Yange Sun, Zhihai Wang, Haiyang Liu, and Chao Du designs an online ensemble paradigm, aiming at combining the best elements of block-based weighting and online processing. An adaptive window algorithm is employed as a change detector. Once a change occurs, a new classifier replaces the worst one of the ensemble. The authors carry out experiments toward both synthetic data streams and real-world data streams to show that the proposed work achieves the best classification accuracy with less memory consumption compared with the previous works.

Considering the fact that variable bandwidth trading can make usage of spectrum more flexible and efficient in spectrum auction, the paper “A Variable Bandwidths Spectrum Auction Mechanism with Performance Guarantee,” authored by Xiaofei Bu, Yu-E Sun, Lina Zhang, He Huang, and Baowei Wang, proposes a truthful auction mechanism which can allocate spectrum with variable bandwidths to maximize social efficiency. A bid-monotone winner determination mechanism with approximation factor of 10 is first designed to decide the winning secondary users in an auction. Then, a channel allocation mechanism is addressed to allocate spectrum to winners without interference. At last, the critical value for each winner is obtained to ensure truthfulness. Both the theoretical analysis and the extensive simulation results show that the proposed auction mechanism is truthful.

The paper “Sports Motion Recognition using MCMR Features based on Interclass Symbolic Distance,” authored by Yu Wei, Libin Jiao, Shenling Wang, Rongfang Bie, Yinfeng Chen, and Dalian Liu, investigates motion recognition in sports training making use of features extracted from distance estimation of different kinds of sensors. For multivariate motion sequence, the authors employ the Max-Correlation and Min-Redundancy strategies to select features extracted through the inter-class distance similarity estimation. In this way, proper features identifying motions in different classes can be effectively screened out. The real-world experiment results demonstrate the effectiveness, accuracy, and low time cost of the proposed work.

To overcome the shortages of global navigation satellite systems, the paper “Correlation-sum-deviation Ranging Method for Vehicular Node based-on IEEE802.11p Short Preamble,” authored by Xuerong Cui, Aaron Gulliver, Hao Zhang, Juan Li, and Chunlei Wu, proposes a correlation-sum-deviation method for ranging using the IEEE 802.11p short preamble to improve the precision of the accurate time of arrival estimation which is important for positioning estimation. It is shown through the simulation results that in both the additive white Gaussian noise channel and the international telecommunications union multipath channel, the proposed method achieves better precision and is less complex compared with the previous techniques, especially in the case when the signal-to-noise ratio is low.

In the paper “Fluidized Bed Agglomeration Diagnose Based on Wavelet Packet Entropy and Gaussian Test,” the authors, Weiguo Lin, Shuochen Wu, Haiyan Wu, Changli Mu, and Yuanhua Qi, propose an agglomeration early warning system to detect the sporadic event of agglomeration. The investigated problem is challenging because of the overheating of the particles in fluidized bed reactors, where sound waves in audible range are employed, and the sampling rate and signal preprocessing approaches are determined based on the energy distribution in the frequency domain and the shift characteristics of the power spectral centroid of the normal and agglomeration signals. The proposed approach is practical and effective, which is verified by the experiments performed on a pilot plant.