Advanced Multimedia and Ubiquitous Engineering

Network traffic classification is elementary to network security and management. Recent research tends to apply machine learning techniques to flow statistical feature based classification methods. The Gaussian Mixture Model (GMM) based on the correlation of flows has exhibited superior classification performance. It also has several important advantages, such as robust to distributional assumptions and adaption to any cluster shape. However, the performance of GMM can be severely affected by the number of clusters. In this paper, we propose the minimum description length (MDL) criterion which can balance the accuracy and complexity of the classification model effectively by evaluating the optimal number of clusters. We establish a new classification model and analyze its performance. A large number of experiments are carried out on two real-world traffic data sets to validate the proposed approach. The results demonstrate the efficiency of our approach.

The literature review suggests that a high quality of system and service, which mobile phone games should provide, appear to be a critical determinant of not only user satisfaction but also the motivation to spend more time and play more games. However, not many past studies in IS have shown much interest in the issues about the effects of the quality of system and service that mobile phone games provide for user satisfaction. Furthermore, there is little research about the relationship between user satisfaction and the motivation to play in the context of mobile phone games. Therefore, this study aims to examine: (1) the effects of the system quality or the service quality of mobile phone games on user satisfaction, (2) the relationship between user satisfaction and motivation to play more games. Structural Equation Modeling (SEM) was employed to analyze data collected through a survey. The results showed that the system quality and the service quality of mobiles games both have a positive impact on user satisfaction. A strong causal relationship between user satisfaction and motivation was also found.

Vehicle ad hoc networks (VANETs) are prone to security risk, especially node misbehavior arising from attacks. In this paper, a certain theory model based model is proposed. We define reputation of a node in VANETs that consist of direct reputation and indirectly reputation. s, direct reputation can be calculated base on certain-factor model according to the records of history communication behaviors. And indirect reputation of a node can be computed based on fuzzy C-means algorithm. The simulation experiment results show that, compared with the baseline method, the proposed methods has better false alarm rate and misdetection rate.

It has been shown that network coding can effectively improve the throughput of multicast communication sessions in directed acyclic graphs, achieving their cut-set capacity bounds. However, network coding is highly susceptible to eavesdropping and pollution attacks in which malicious nodes attacks cannot be prevented. In this paper, we propose several ways to enhance the security and reliability of random networking coding transmission. Schemes which encrypt the random coefficients and the coding payload could ensure the information has a relatively low probability to be cracked. We also implement the reliable transmission by forward error-correction. It has been shown that random network coding with encryption and forward error-correction can help achieve provably good overall security and reliability performance.

At present elderly people seem to be one of the rapidly growing population groups in the developed countries. This trend of aging population causes additional problems such as increased costs on the treatment and care about this elderly people. Therefore there is ongoing effort to extend the active age of this group of people in order to enable them to stay economically and socially independent. And current technological devices and services can assist them in this process. The purpose of this article is to discuss an attitude of elderly people towards mobile technological devices and benefits and limitations of mobile applications for this group of people. The author used a method of literature review of available sources exploring research studies focused on mobile devices and applications for elderly people in the acknowledged databases and a method of comparison and evaluation of their findings.

Geographic routing (GR) is designed for forwarding packets within a specific geographic region, and it enjoys the advantages of its scalability and simplicity. GR can provide a promising solution for packet delivering in next generation wireless network, and has gained much research attention in the areas of Wireless Sensor Networks (WSNs), Vehicular Ad Hoc Networks (VANETs) and Mobile Ad hoc Networks (MANETs). However, it suffers from communication holes or voids in the network areas due to network dynamics or random deployments. Most of current void handling schemes use local or whole topology information. In this paper, we propose a heuristic geographic routing (HGR) to avoid holes over a few iterations by using only local geographic information. Through simulation and analysis, we find that HGR has an outstanding performance in the respect of void bypassing. In addition, this algorithm keeps simplicity and has low communication overhead.

Because a wide range of professionals utilize Social Network Service (SNS), the SNS users have recently required an expert recommendation service to enable users to perform both cooperation and technical communication with experts. A content-boosted collaborative filtering (CBCF) provides various prediction algorithms which support effective recommendations. However, the CBCF cannot calculates the similarity of items (or users) when the calculation condition is not clearly provided. To solve the problem, we propose a content-aware hybrid collaborative filtering scheme for expert recommendation in SNSs. Finally, we show from a performance analysis that our scheme outperforms the existing method in terms of recommendation accuracy.

The problem of automated theorem finding is one of 33 basic research problems in automated reasoning which was originally proposed by Wos. The problem is still an open problem until now. To solve the problem, a systematic methodology with forward reasoning based on strong relevant logic has been proposed. This paper presents a case study of automated theorem finding in Tarski’s Geometry to show the generality of the methodology.

An important trend in the area of IT at present are mobile technologies and mCommerce. They can be also exploited by seniors who in developed countries are becoming to be perceived as a new group of potential customers and users in all areas of human activities, including ICT. The aim of this article is to describe a potential of mCommerce for seniors whose number in the developed countries will be increasingly rising. The methods used for this study include a retrospective analysis of available sources in the area of the use of ICT by seniors, a data analysis from the world’s databases (Eurostat, WHO) and an analysis of the external environment of ICT sector based on the previous studies. The findings show that the main benefits for the seniors in the use of mCommerce seem to be direct and fast accessibility to information, comfortable electronic payment transactions, cuts of variable costs, and thus the overall improvement of quality of life. On the contrary, the main obstacles include security and trust issues, mobile device limitations and marketing challenges.

The latest version of programming language Ada, Ada 2012, has introduced the concept of contract-based programming (CBP) and became the first internationally standardized programming language to include CBP as an intrinsic feature of the language. CBP can strictly stipulate and assure the correctness of programs to enhance the reliability and security of safety-critical systems, due to terrible design and/or programming practice, there is an issue that it might obstruct some other factors of the software quality. Therefore, it is essential to implement a supporting environment for CBP with Ada 2012 in order to not only retain reliability by using CBP, but also avoid taking interference to other factors of software quality. Until now, there is no report for proposing supporting environments or tools for CBP, while most studies focus on how to check the conditions of contracts, i.e., what contracts should do for software engineering activities. To support CBP with Ada 2012, this paper analyzes the issues that CBP disturbs some other factors of software quality, proposes methods to avoid the issues, and shows a supporting environment for CBP in Ada 2012 programs.

The development of ICT goes hand in hand with a rapid growth of company data volume. Much of the data contains valuable information, which can be used for the company’s further development. The aim of this paper is to analyze the use of business intelligence in European business sector. Technological and economic aspects of the use of business intelligence will be described. The attention will also be focused on mobile BI. Moreover, in the end they will be summarized in the SWOT analysis. The used methods are analysis of the external environment and the subsequent SWOT (Strengths, Weaknesses, Opportunities, Threats) analysis. The external environment involves forces outside the ICT (information and communication technologies) sector in Europe that can potentially influence the use of business intelligence. The analysis shows that, given the corporate internet facilities in Europe, there is great potential for its use that is not currently used.

The mainstream risk decision making models, which follow the maximization rule, are based on the unbounded rationality hypothesis. However, the Equate-to-Differentiate model, which is a heuristic framework and follows the satisfaction rule, challenges these dominating models by proposing a bounded rationality approach. This study contributes to research on dual-process techniques by applying both models for explaining inference strategies that are used in individuals predicting and reasoning behaviors. To test its empirical validity, we demonstrate evidence that our model can account for real practical choices which are anomalies or paradoxes from the point of view of many other risk decision making models.

The knowledge checking service (KCS) is critical for safeguarding the quality of knowledge. In organizations, there are many knowledge checking services with similar functions but different qualities, in order to select the best KCS, the method which combines grey related analysis (GRA) with data envelope analyzes (DEA) with incomplete weight information in fuzzy environment is proposed. In the method, the evaluation information is given linguistically and processed by the 2-tuple linguistic model and GRA. The DEA model is constructed to derive the precise weight information. Afterwards, by solving DEA model, the order of the alternatives can be derived. In the method, the combination of the advantages of DEA and grey correlation analysis leads to the avoidance of subjectivity and make the selection more accurately. The applicability of the proposed method is validated by an example.

Secure range query processing algorithms have been studied as the range query can be used as a baseline technique in various fields. However, when processing a range query, the existing methods fail to hide the data access patterns which can be used to derive the actual data items and the private information of a querying issuer. The problem is that the data access patterns can be exposed even though the data and query are encrypted. So, in this paper we propose a new range query processing algorithm on the encrypted database. Our method conceals the data access patterns while supporting efficient query processing by using our proposed encrypted index search scheme. Through the performance analysis, we show that the proposed range query processing algorithm can efficiently process a query while hiding the data access patterns.

Internet of Things (IoT) has become a popular paradigm by digitizing our physical world, bringing appealing level of conveniences to community. Besides that consumers embrace potential IoT benefits to them, they are much more concerned about the security and privacy of their data. Since, the increased connectivity between devices and the internet in IoT presents security and privacy related challenges, and the data breach of sensitive personal information can be exploited to harm the consumers; the trust of these consumers on the IoT can be tampered. The success of IoT ultimately depends upon the consumer perceptions about security and privacy in IoT and the level of digital trust of its consumers. Since trust has been an important element to influence consumer’s behavior, thus, it can be stimulated by increased consumer perceptions of both security and privacy of IoT. In this paper we contribute to identify the evolving features of IoT and identify important security and privacy requirements from consumer perspectives. We propose a conceptual security and privacy model and highlighting important threats and challenges on various stages of this model.

On the surface of the quad flat non-lead (QFN) dark-filed images, noise pixels (including textures produced in the molding process) obstruct the defect inspection. To extract defects from QFN surface, a novel method based on texture detection and region of interest selection is proposed. Firstly, a QFN texture direction detector is proposed. Secondly, multilevel thresholding method is used to segment QFN images. Thirdly, according to the image level, the bright defects images and the dark defect images are obtained. Then, the region of interest selection method is applied to reserving defects regions and removing QFN textures and noise pixels. Finally, our method extracts defects by combining the bright and dark defects image. The experiments show that the proposed method can extract defects efficiently.

Conventional pattern classification aims to improve classification accuracy for the whole dataset. In the time of Big Data, however, there are circumstances in which people may take interest only in those typical instances and other issues like scalability and efficiency take priority. Keeping these issues in mind, in this paper, we revisit the perceptron algorithm. While it is a linear model, we show that with proper objective functions, it can be transformed to a probabilistic learner. The evaluation is carried out with the well known Pima diabetes database. The experimental results indicate that the perceptron algorithm is comparable to other sophisticated sophisticated algorithms in terms of the criteria discussed in this paper.

In particle-based fluid simulations, viscoelastic materials require yield stress for material deformation. We propose an ideal viscoelastic material yield condition developed by modifying the Tresca’s condition that can be easily approximated using the difference between the maximum and minimum principal stress, unlike the von Mises condition for which the forces in numerous directions applied to an object should be calculated. The proposed ideal viscoelastic material yield condition assumes the area of the object deformed due to the forces applied to it as the principal stress based on the Tresca yield condition. Using this method, the process through which a viscoelastic material explodes because it cannot endure the critical stress when its interface decreases beyond the ideal yield condition can be realistically expressed.

This paper presents a collaborative strategy of the distributed RFID sensor clusters for object association and identification. The process that converts the group associations to the single associations is also described. The basic association mechanism utilizes two homographic regions that model the RFID fluctuations. The sources of potential generation of false associations are discussed and the techniques for eliminating them with sensor collaboration are presented. The problem with the information inconsistency and its temporal propagation in the large-scale is described, and the mechanism for detecting and correcting such an inconsistency is proposed. The association performance of the proposed strategy is simulated with various parameters.

The growing concern about IAQ has accelerated the development of small, low cost Indoor Air Quality (IAQ) monitoring systems. These are capable of monitoring various indoor air pollutants in real-time. However, since most IAQ monitoring systems present the sensed value of the corresponding pollutant as a number, when we read the sensor values it is difficult for not experts but normal users to identify how polluted the air is or what is the pollution criteria of each pollutant. Therefore, for ambient air quality, a comprehensive Air Quality Index (AQI) is already being used in a number of nations to present the polluted level of the air, and thereby people can identify the current status of air quality easily. Nevertheless, since the index calculation is based on ambient air pollutants, and the index presents a section mean of sensed values collected for 1 or 24 h, the AQI used for presenting ambient air quality is not suitable for IAQ representation. Therefore, in this paper, we perform an experiment that applies the AQI directly to IAQ monitoring system that we developed, and derive some problems through the result analysis.

In this paper, to deal with stream big data processing issue, we design and implement a big data analysis system using Spark which is an In-memory cluster computing framework. Spark is provided by ASF (Apache Software Foundation) open-source community, and is regarded as a next-generation high performance cluster computing technology. From the performance evaluation of the proposed system, we can see that Spark is 20+ times faster than conventional Mapreduce-based Hive SQL in terms of the response time. According to these results, we can confirm that the proposed system can be applied to solve the soft real-time big data analysis jobs for sensor data generated in a smart factory.

We researched the performance of AODV routing protocol based on NS-2 in Ad Hoc networks on the Linux platform. When node’s moving speed is changing, the typical Random Way Point mobility model is analyzed by using NS-2. And when the vehicle’s running speed is changing, the IDM_LC model is analyzed by using VanetMobiSim and NS-2. The performance of AODV routing protocol is evaluated on the basis of End-to-End Delay, Packet Loss Rate and Throughput. Finally, we compared the performances of AODV routing protocol as the node’s moving speed is changing on these models. The results show that different mobility models have different effects on the network performance of AODV routing protocol.

In this paper, we develop a touch detection system using the Kinect-projector system. In our system, the touch area is first determined using a user’s assistance with the Kinect provided skeleton data, and the touch is then detected within only the touch area to simplify and speed-up the process. In particular, a precise touch location is found considering the user’s touch behavior and depth characteristics. Experimental results demonstrate that the proposed system can be used for touch sensing on planar walls without requiring any active touch sensors.

As the importance of automotive ECU (Electronic Control Unit) and its software increase, the systematic testing method is applied to them. However, it takes a lot of time to localize the faults because the developers have not been enough information which can be used for debugging by the nature of the test process for the automotive software. In this paper, we propose a method to reduce the fault-suspicious region in the memory by utilizing the memory map and the correlation between the inputs and the update information. As a preliminary result, we confirmed that the fault-suspicious region is reduced to 17.42(%) of the memory size by using the proposed method.

The mobile context-aware service is performed based on the collected data from the various sensors. Additional power is consumed by the operations of the sensors which lead to the reduced battery life. So it is necessary to manage the sensors efficiently for sustainable mobile context-aware service. Among the sensors, GPS is well known as a noticeable battery hog affected by the surrounding environment. The power of the mobile device is consumed unnecessarily due to the operation of the GPS in the place where the locational data is not updated. Therefore, the efficient management of the GPS is required and analysis of the power consumption of the GPS has to be preceded. However, in the existing studies, the GPS has been analyzed with course-grained models such as ON/OFF or enabled/disabled so it is hard to get information enough to identify what the state is causing unwanted power consumption of the GSP. In this paper, we propose a technique for analyzing the power consumption of the mobile device according to fine-grained states of GPS using location APIs.

A technique to communicate with a smart sensor device using the Web is currently being developed. When the user approaches the device, like a beacon, the device sends a URL to the built-in sensor to the user’s smartphone. With this URL, the user can communicate with the smart sensor device through a Web interface. When communicating with a beacon using a Web interface, encryption is required for secure communication between the Web browser and beacon. In this paper, we propose secure communication using a symmetric key distribution. Our method prevents malicious applications from eavesdropping on messages and replaying a message. In addition, we evaluated the execution time.

Virtual reality (VR) with head-mounted display (HMD) device provides an immersive experience for novel multimedia applications. This paper develops an interactive virtual reality system with a wireless HMD to enable a natural VR operation interface. In a server, the system utilizes the Kinect as a motion detection device to estimate VR user’s location and gesture information in real time. Through a WiFi network, the user’s information is transferred to the HMD as a client, where the user controls an avatar following his motion. The controlled avatar interacts with the virtual environment in real time. The proposed system is implemented using a Samsung Gear VR, Kinect 2.0 and Unity3D environment. This system is compatible with serious games, virtual and physical collaboration, natural user interfaces, and other multimedia applications.

As information technology service matures to an even higher level and as sensor technology becomes embedded on person’s life, the needs to support and satisfy instant personal learning activity on own life without the limitation of time, space, and additional complement have become a vital point to everyday lives. Especially smartphone users are increased in a short-term lesson and exercise, and instant and short learning contents on mobile learning model. And programing subjects for distance learning are limitedly supported, since virtual practice environment is not appropriate for tutoring service. Programming practices lesson has frequent program errors and needs error correction by oneself or by a tutor. But on distance learning, it is almost impossible for a learner to be guided or correct lesson from a tutor. In this paper, well-formed programming practice scenarios, by that a learner could follow and learn program practice lesson, are supported and delivered by small-bite size learning contents. For well-formed programming practice scenarios, small bite learning contents are 10–15 min length and consists of lecture video, synchronized lecture slides video with lecture video, digital textbook, sample program execution drills, and modifiable program execution drills. Programming Exercise, a mobile learning app, provides learners with interactions between the learners and the contents by program practice.

In recent years, the technological advancements have led to a deluge of data from distinctive domains and the need for development of solutions based on parallel and distributed computing has still long way to go. That is why, the research and development of massive computing frameworks is continuously growing. At this particular stage, highlighting a potential research area along with key insights could be an asset for researchers in the field. Therefore, this paper explores one of the emerging distributed computing frameworks, Apache Hama. It is a Top Level Project under the Apache Software Foundation, based on bulk synchronous parallel model. We present an unbiased and critical interrogation session about Apache Hama and conclude research directions in order to assist interested researchers.

This paper proposes the interactive lecture system based on mixed reality using a transparent display that enables interaction with the users watching the display. This system unlike existing remote video lectures, will allow students to participate progressively in the lectures by enabling the interaction between the instructor and students. Furthermore, this system also improves the reality in such a way that, during the visualization of the real-time video lectures on the transparent display, it seems that the instructor is in the same place with the students. This paper experiments and implements the proposed system based on the transparent display, 2D camera, and 3D sensors.

For scientific educational purposes, many types of chemical experiments are conducted in academic institutions. However, these experiments in real classroom environments can be quite dangerous and can potentially harm students. To instruct students in conducting experiments safely, virtual chemical experiment applications were developed. However, existing virtual science experiment applications are mostly limited to having only one participant. To solve this limit, this paper describes the design of an interaction engine, enabling interaction among multiple users and implements the virtual science experiment application based on that interaction engine.

Despite the extensive research into 3D eye-tracking methods, such methods remain dependent on many additional factors such as the processing time, pose, illumination, image resolution, and calibration procedure. In this paper, we propose a 3D eye-tracking method using the HD face model of Kinect v2. Because the proposed method uses accurate 3D ocular feature positions and a 3D human eye scheme, it can track an eye gaze position more accurately and promptly than previous methods. In an image captured using a Kinect v2, the two eye-corner points of one eye are obtained using the device’s high-definition face model. The 3D rotational center of the eyeball is estimated based on these two eye-corner points. After the center of the iris is obtained, the 3D gaze vector that passes through the rotational center and the center of the iris is defined. Finally, the intersection point between the 3D gaze vector and the actual display plane is calculated and transformed into pixel coordinates as the gaze position. Angle kappa, which is the gap between the actual gaze vector and the pupillary vector, is compensated through a user-dependent calibration. Experiment results show that the gaze estimation error was an average of 47 pixels from the desired position.

In this paper, we introduce a fault-tolerant inVANETs-based intersection control algorithm which relies on vehicle-to-vehicle or vehicle-to-infrastructure communications to control the traffic and grant the privilege to cross the intersection. The primary-backup approach is adopted in this work, where a backup controller is added to monitor the primary controller and to take on once it discovers that it has failed. The proposed solution guarantees that the safety, liveness, and fairness properties are satisfied all the time.

E-questionnaire, e-testing, and e-voting are the essential ingredients of modern communities as the methods for a group to express a choice, a preference, or an opinion by an e-paper. Many kinds of e-questionnaire, e-testing, and e-voting systems are implemented to provide e-questionnaire, e-testing, and e-voting services on the Internet. However, there is a gap manifested in difficult communications among questioners, developers, and systems. To cover the gap, this paper proposes QSL, the first specification language with a standardized, consistent, and exhaustive list of requirements for specifying various e-questionnaire, e-testing, and e-voting systems such that the specifications can be used as the premise of automatically generating e-questionnaire, e-testing, and e-voting systems. This paper also presents QSL structure satisfying stability and extensibility, shows various QSL applications for providing convenient QSL services to questioner and developer.

Knowledge about attacks is a necessary foundation for security analysis of information systems or cryptographic protocols. Current security verification methods for improving the security of target systems or the soundness of cryptographic protocols has limitations because they are all based on the assumptions from known attacks, while the attackers are trying every possible attacks against the information systems. Once a new-style attack was found by adversaries earlier, it would bring severe loss to the target systems. Therefore, it is essential to understand and take measures against new attacks previously. A new method has been proposed for predicting new attacks, but it lacks experimental results to prove its effectiveness. This paper confirms the effectiveness of the proposed method by a rediscovery experiment that shows several known attacks on cryptographic protocols rediscovered successfully. The paper also shows issues of the approach for predicting new attacks.

This paper proposes a dynamic traffic data visualization system with OpenStreetMap (OSM). The system connects server database to client web browser by a Transmission Control Protocol/Internet Protocol (TCP/IP) to access remote traffic data. In order to reduce the computation consumption of the server, the traffic estimation and prediction from large datasets is analyzed in the client. We implement a Graphic Processing Unit (GPU) programming technology to implement the data mining process in parallel for real-time approach. To provide an intuitive interface to the users, the system renders the data mining results with the OSM mid-ware, which provides geographic data of the world.

Human face recognition performances usually drops heavily due to pose variation and other factors. The representative deep learning method Deep Belief Network (DBN) has been proven to be an effective method to extract information-rich features of face image for recognition. However the DBN usually ignore the local features of image which are proven to be important for face recognition. Hence, this paper proposed a novel approach combined with local feature Center-Symmetric Local Binary Pattern (CS-LBP) and DBN. CS-LBP is applied to extract local texture features of face image. Then the extracted features are used as the input of Deep Belief Network instead of face image. The network structure and parameters are trained to obtain the final network model for recognition. A large amount of experiments are conducted on the ORL face database, and the experimental results show that compared with LBP, LBP combined with DBN and DBN, the proposed method has a significant improvement on recognition rates and can be a feasible way to combat with pose variation.

When searching for historical topics on Web search engines, the query results are not displayed in chronological order of documents content. Hence the user has to manually navigate through the search result in order to plot the documents on a time axis. If documents can be sorted in contextual chronological order automatically, it would have various practical applications. To test if this concept is feasible, we analyzed the Annals of the Joseon Dynasty, which is a compilation of daily journals spanning six hundred years, and applied various approaches of machine learning algorithms to estimate the approximate temporal information of historical documents written in modern period. Our experiment showed the accuracy as high as 64 %, suggesting that estimating temporal information based on document text is feasible.

In LTE-based mobile communication systems, a Tracking Area Code (TAC) defines a group of cells for a paging area. In mobile networks, the paging performance is a critical factor to be considered, since it may give large impacts on paging response time as well as paging traffic load in the mobile system. This paper proposes a mobility-aware TAC configuration scheme to increase the paging success rate in mobile communication systems. We first construct an optimization model for TAC configuration by considering the mobility (handover) patterns of mobile users as well as the TAC size and the capacity of paging traffic. Then, we propose the tow algorithms for TAC configuration to maximize the paging success rate, while some constraints are satisfied. From the performance analysis with real traffic data of SK Telecom in Korea, we can see that the proposed TAC configuration provides larger paging success rates than the existing TAC configuration.

In this paper, a two-stage estimation filter is proposed to consider both nominal system and temporarily uncertain system by applying infinite memory structure (IMS) and finite memory structure (FMS) estimation filters selectively. One of two filtered estimates is selected as the valid estimate according to presence or absence of uncertainty. A detection rule is developed to indicate presence or absence of uncertainty and select the valid filtered estimate from IMS and FMS filtered estimates. The detection rule consists of uncertainty presence and absence detections. Two kinds of test variables for the detection rule are defined using the chi-squared distribution with one degree of freedom. Finally, to verify the proposed two-stage estimation filter, computer simulations are performed. Simulation results show that the proposed two-stage estimation filter works well for both nominal system and temporarily uncertain system.

The WBSN is a network environment in which various types of bio-signals generated directly or indirectly inside and outside the body are measured and processed for transmission to monitor the condition of the patient. The conventional DTD (Decrease of Transmission Delay)—MAC protocol transmits general and emergency data without any classification. As a result, the average delay and packet loss rate increase. Hence, in order to mitigate this performance degradation, we propose two types of adaptive MAC protocols to deal with emergency data separately in the system. The first proposed protocol reduces the delay by prioritizing emergency data over other general data by sending them first. The second proposed protocol applies the maximum delay requirement to emergency data to reduce the packet loss of both general and urgent data packets at the same time.

In this paper, we propose a buffer-aided relay selection scheme with primary sensing ability in an underlay cognitive radio network. The proposed relay selection scheme is evaluated in terms of outage probability, compared with the conventional max-min and max-ratio relay selection schemes, through simulations. The results show that the proposed scheme significantly improves the outage performance with a low primary activity and a short sensing period.

Persons with visual impairment have low and tunnel vision, and the persons with motor impairments have muscle weakness and hand tremor. Selecting an object requires constant use of hands and fingers. This increases fatigue and eventually causes muscle damage and strain. As a result, visual and motor impaired persons give up selecting the target object. Traditional cursors are designed to aid quick selection and to reduce error. However, it is insufficient to reduce the fatigue on hands and fingers. In this paper, we suggest a cursor using limited range of motion, CLROM, for persons with visual and motor impairments. CLROM reduces muscle fatigue and prevents muscle damage by limiting the movements of the joints in outer range and middle range. CLROM prevents the losing of the cursor position and overshooting by limiting the motion range to the area cursor. Consequently, CLROM helps to acquire the target object easily and reduces error and muscle fatigue.

In recent years, fusion studies such as context-aware, multimedia contents and cloud computing in information technology (IT) have been on the rise through the technological development of hardware and software. The automatic context-aware technology using multimedia image data requires high computation. Cloud computing has played a role in meeting the requirements of high computation. The automatic context-aware technology utilizing previously developed multimedia image data has lacked user-defined semantic inference capabilities considerably. This paper proposes a Semantic Image Processing Mechanism for Automatic Context-Aware (SIPM-ACA) based on cloud computing. Semantic inference is done through user-created multimedia contents images. Image’s semantics are verified by analyzing texts created by other users. Content-Based Image Retrieval (CBIR) is utilized to find the relationship between image similarities. Through this, proactive context-awareness according to user’s context can be inferenced.

A 2D barcode region localization system for the automatic inspection of logistics objects has been developed. For the successful 2D barcode localization, frequency of the pixel distribution within average 2D barcodes is modeled and the average model of 2D barcode is combined with the corner features to localize the objects having high possibility of 2D barcode candidates. An automatic 2D barcode localization software was developed with frequency and corner features and we tested our system on real camera images of several popular 2D barcodes. It improves on runtime of our previous method.

Cloud computing is a widely used technology in software industry. In cloud computing, virtualization technique is used to provide most of the services. Mostly cloud providers use Virtual machines to satisfy the user requests. Efficient scheduling of virtual machines is an important job in cloud computing. In this paper, we study Eucalyptus, Open Nebula, and OpenStack cloud virtual machine scheduling algorithms. Eucalyptus cloud uses Round Robin and Greedy virtual machine scheduling algorithms. Open Nebula uses match making scheduling algorithm and filter scheduling algorithm is used for virtual machine scheduling in OpenStack. Round Robin, Greedy, Match making, and Filter scheduling algorithms are compared in this paper.

A variety of new security technology has emerged in the mobile security area recently, especially domain isolation technique is widely used, such as TrustZone, Samsung KNOX, etc. By storing user sensitive information and business data in a secure domain, which is isolated from normal domain, may not be exposed to unexpected security accident or unauthorized access. When the security incidents occurred on these devices, it might be impossible to collect data from secure domain, because common forensic tools cannot be accessed in isolated domain. Therefore, it is necessary to research data collection techniques on the device based on domain separation technology. This paper discusses data collection techniques in the secure domain applied by mobile hypervisor-based separation technology.

This work presents an optimized method to simulate the modeling of photovoltaic (PV) modules with measured data of PV array. The current-voltage (I-V) characteristics are estimated via adaptive neuro-fuzzy inference system (ANFIS). The proposed ANFIS method takes advantages of no need of internal parameters of PV model and can achieve a more accurate estimation of PV characteristics. By compared with Villalva’s model, radial basis function neural networks (RBFNN) and support vector machine (SVM) method, the results predicted by the proposed ANFIS approach show the best estimation performance in terms of root mean squared error (RMSE), mean absolute percentage error (MAPE) and coefficient of determination (R2).

Sparse representations of signals have become an important tool in computer vision. In this paper, we propose a non-linear non-negative sparse representation model: NNK-KSVD. In the sparse coding stage, a non-linear update rule is proposed to obtain the sparse matrix. In the dictionary learning stage, the proposed model extended the kernel KSVD by embedding the non-negative sparse coding. The proposed non-negative kernel sparse representation model was evaluated on two public image datasets for image retrieval, promising image retrieval performance was obtained.

A content-based video retrieval via visual feature pooling is proposed in this paper. Since these visual words represent local features extracted from frame images, spatio-temporal constrains are applied to solve the ambiguity of the model towards effective retrieval of semantic video clips. Both shot level and segment level processing are employed, and the latter is found more robust in dealing with complex scenes where accurate video segmentation may fail. Our experimental results have shown that the constrained scheme help to improve 5 % average matching accuracy. In addition, it suggests that summarized videos at 25–30 % of original size can still maintain a viewing quality of 70–80 % towards fast content delivery.

Support Vector Machine (SVM) classifiers are binary classifiers in nature, which have to be coupled/assembled to solve multi-class problems. One-Versus-Rest (1-v-r) is a fast and accurate method for SVM multiclass classification. This paper investigates the effect of output grouping on multiclass classification with SVM and offers an even faster version of 1-v-r based on our output grouping algorithm.

Quantitative information of maximum power point (MPP) is crucial for controlling and optimizing the output power of photovoltaic (PV) modules. However, it is difficult to obtain the voltage at MPP through direct measurements. A novel approach of radial basis function neural network (RBFNN) is proposed to achieve maximum power point estimation in this study. The proposed method has the capability of determining the MPP of PV arrays directly from the measured current–voltage data of PV modules, and takes advantages of no need of internal parameters of PV model. The experimental results show that the proposed approach can obtain the optimal power output in high accuracy.

In time series forecasting, noise can have a cumulative effect on the prediction of future values thus impacting the accuracy of the model. A common method of machine learning in time series problems is to provide a number of past output values in the series so it can learn to predict the next value, however, other modes of time series forecasting also include one or more input series. This enables the application of the proposed technique in this study to provide additional meta-information to the model to guide learning and improve the prediction performance of the model. We identified the components of two time series datasets using empirical mode decomposition and trained a non-linear autoregressive exogenous model to compare its performance with the traditional approach. Two methods for processing the signal components for noise reduction were proposed and the result from the summed combination significantly outperforms the traditional technique.

Android platform applications are the most dominant technologies in the mobile markets. These applications have a wide range of functionalities such as Game, Business, Education, Entertainment, Shopping, Travel and Weather etc. However, most of these applications have a performance problem in their Graphical User Interfaces (GUIs) since the mobile application GUI testing is daunting and too expensive. In this study, we investigated the severity of the Android applications GUI performance problem by examining a sample of freely downloadable applications from Google Store and analyzed performance issues of the applications.

This paper proposes a feature-vector for characterizing performance test cases. The feature-vector is composed of six attributes such as use of thread, measuring elapsed time and counting successful and failed number of test cases. In order to identify the feature vector, we thoroughly examined the test cases from five open source projects and extracted performance cases. After then, we established the common feature vector discovered from the performance test cases, and analyzed distribution of the feature vector in the performance as well as general test cases to show the validity of the feature-vector.

Describing semantic attributes of clothing is an important technique for many applications. In this paper, we investigate problems of predicting the high-level semantic attributes by making use of features learned from middle-level semantic attributes and analyzing the hierarchical relations among the semantic attributes. In order to crop clothing regions from original images under various conditions, a modified Faster-RCNN [1] is implemented. We propose a novel hierarchical semantic tree-structure deep neural network to model the relations between middle-level and high-level semantic attributes, which improves the prediction accuracies of high-level semantic attributes. A large number of experiments are performed to verify different contributions of middle-level semantic features to the high-level semantic attributes.

Natural phenomena simulation has attracted a spurt of research attention and interest in both computer graphics and virtual reality technology domains. Water is one of the most common phenomena in real world. Acquisition and modeling of different water movements have become one of the important research topics in recent years. In this paper, we design an acquisition system for capturing dynamic water surface. We also present an image-based method for water surface reconstruction. The goal of water surface reconstruction is to compute 3D coordinates and normals of dynamic water surface.

Recently, a novel reversible data hiding scheme is proposed for contrast enhancement by Wu (IEEE Signal Process Lett 22.1:81–85, 2015). Instead of pursuing the traditional high PSNR value, he designs the message embedding algorithm to enhance the contrast of the host image. In this paper, an extended scheme is proposed to not only adaptively enhance the contrast of the image, but also to keep the PSNR value high meanwhile. Firstly, the original host image is divided into non-overlapping blocks, such that the local contrast of the image can be enhanced adaptively. Secondly, we classify the pixels of each block into two sets, the “referenced” set and the “embedded” set, and then processing them alternatively such that additional side information is eliminated. Experimental results demonstrate that our proposed algorithm achieves increased local visual quality and performs better than Wu et al.’s scheme with keeping image’s PSNR high as criterion for RDH.

In this paper, we introduced the state-of-the-art techniques for digital image forensic authentication. A novel system was presented which consisted of the advanced tools for digital image forensic authentication. The tools offered in the system had the functions including the image file header analysis, device identification, re-compression analysis, copy-paste detection, resampling analysis, etc. The system performance has been tested under practical cases of forensic image examination. The experimental data and results have proven the advancement of the proposed system which could be widely applied in forensic image examination and other forensic applications.

In this paper we present a method to generate high-quality hard shadows in real-time based on geometry shadow map. The method focuses on addressing the aliasing artifacts due to the large number of overlapping primitives which cannot be stored sufficiently and yields error of depth reconstruction. We call it Covered Primitives Parallel Detection algorithm (CPPD). In CPPD, a parallel detection algorithm according to the texel edges is proposed without increasing storage of different triangles. Experiments show that CPPD algorithm can improve the accuracy of depth reconstruction and reduce the jagged shadows effectively.

Recently, the amount of data is rapidly increasing and thus MapReduce has attracted much interest as a new paradigm for such data-intensive applications. Similarity join is an essential operation for data analytics, including record linkage, near duplicate detection, document clustering. However, the performance of MapReduce is limited when applied on complex data analytical task involving joins of multiple datasets. Hence, workload-aware data partitioning techniques are required, which ensure the balance of computation of each machine. In this paper, we propose a similarity join algorithm using MapReduce that provides scalability and high performance by using grid-based data mapping technique for joining datasets. From the experiment analysis, we prove that our algorithm outperforms the existing algorithm under various data size and similarity thresholds.

Recently, due to the activation of internet networks and development of wireless systems that can establish a network independently, the technical fields where the game theory plays an important role are increasing continuously, solving the decision-making problems in the computer networks. Meanwhile, the development speed of Smart Grids, Micro Grids and IoT-related technologies is also increasing backed by the major IT companies to provide more efficient power management and control services. The game theory used in these technologies allow consumers to check their power usage patterns for better consumption behavior. The power companies use the game-theoretic approaches in their applications and one such example is the ‘Demand Response’ technology for the Smart Grids, which informs hourly or periodic dynamic prices to the power users. Future energy shortage is a global problem and an inevitable consequence for the Republic of Korea as well. Without the ground-breaking power sources and technologies being available, it is expected that the power companies will offer more segmented power prices depending on the time zones and seasons to save power or to reduce the energy consumption rate. Thus, in this paper, we present a game theory-based pricing approach in a Smart Grid and a hybrid AMI which considers the weather conditions. With this AMI, the future, spot and option prices will be determined based on the decision-making process and by applying the process to the Smart Meter System, an efficient monitoring and management will become possible.

The network traffic generated by humans and various devices is one of the most important data sources in network forensics. The main challenge in investigating and collecting evidence in network traffic is handling the huge amounts of data streams caused by the rapid growth of network bandwidth and applications, as well as preserving the useful information for further analysis. HTTP, as the most popular protocol on the Internet, is usually exploited to carry malware and evasive attacks besides the normal services. In this paper, we study how malware and network attacks in real-world exploit HTTP to hide their malicious activities and present an Evasive Network Attack Forensic System (ENAFS), which is able to effectively discover evasive network attacks on HTTP and integrally draw attack the samples and their metadata for further analysis. We believe that our work will benefit the research in the network forensics field in the future.

With the explosive growth of mobile terminal access to the Network and the shortage of IPv4, the Network Address Translation (NAT) technology has become more and more widely used. The technology not only provides users with convenient access to the Internet, but also brings trouble to network operators and regulatory authorities. This system NAT detection using NetFlow data, is often used for monitoring and forensics analysis in large networks. In the paper, in order to detect NAT devices, an Out-in Activity Degree method based on network behavior is proposed. Our approach works completely passively and is based on NetFlow data only. Our approach gets accuracy of 91.2 % in real large-scale network for a long time.

The movement of human action recognition technology is the key to human-computer interaction. For the movement of human action recognition problem, this paper has studied the theoretical basis of hidden Markov models including their mathematical background, model definition and hidden Markov model (HMM). After that, we have built the establishment of human action on hidden Markov models and train the model parameters. And this model can effectively target human action classification. Compared with conventional hidden Markov model, the method proposed in this paper to solve the movement of human action recognition problem attempts to establish a model of training data according to the characteristics of human action itself. And according to this, the complex problem is decomposed, thus reducing the computational complexity, to the practical applications to improve system performance results. Through the experiment in the real environment, the experiment show that the model in the practical application can be identification of the different body movement actions by observing human action sequence, matching identification and classification process.

Since image processing software is widely used to tamper or embed data into JPEG images, the forensics of tampered JPEG images now plays a considerable important role. However, most existing forensics methods that use binary classification can hardly deal with multiclass image forensics problems properly under network environments. In this paper, we propose a hierarchical forensics scheme against multiple heterogeneous JPEG images. We introduce a compression identifier based on Markov model of DCT coefficients as the first hierarchical section and then develop a tampering detection and steganalyzer separately as the second phase. We conduct a series of experiments to testify the validity of the proposed method.

Today, with the increasing development of science and technology, social network software is accepted by more and more people. For most of the people, exchanging information and instant messaging are the function of social software. That makes people's lives very convenient. But it has also become the powerful tool for criminal activities. Therefore, the evidence of social network software has also become an effective evidence to prove the crime.

Recently, user-centered user interface design principles for PCs have been presented, such as design consistency or design simplicity. However, those principles are generally based on design specialists’ experiences, and thus there are relatively few studies that have explored the design principles based on a theoretical foundation. Therefore, the need for empirical studies on the heuristic GUI design principles has been gradually increasing. This study was designed to identify the effects of the heuristic GUI design principles presented in the environment of PCs on the system users’ convenience, i.e., particularly on information access in the current mobile computing environment. The aims of this study are to verify the influence of design consistency and simplicity employed for GUIs of mobile applications on the users’ understandability of GUIs; and furthermore, to identify the effects of GUI understandability on the accessibility to information via mobile applications of users. Data were collected through a survey and Structural Equation Modeling (SEM) was employed to analyze the data. The results found that in the environment of mobile application, there was a verification of: the influence of design consistency and design simplicity on the users’ GUI understandability of mobile applications; and furthermore, the resulting effects of that GUI understandability on the accessibility to information of users.

Recent development in information technology has changed the distribution system for agri-food products. Along with the development in agri-food products, it became necessary to develop an e-commerce system that allows customers to search information about and purchase agri-food products. For the purpose, this study　developed　a virtual store where customers can purchase agri-food products using smartphones and mobile devices. To development of mobile based virtual store with agri-food ICT convergence—integration (1) implement of agri-food products display, (2) construction of code interface for devices and connected products, (3) development of Service Platform and (4) implementation of 3D content creation system.

This paper analyzes different disease forms among various administrative areas. It is able to help healthcare authorities make quick and exact decisions in the allocation of medical resources and health services. To solve this problem, we exploit four different clustering algorithms,including Hierarchical Agglomerative Clustering (HAC), K-Means, Density-Based Clustering (DBSCAN), Gaussian Mixed Model (GMM). Furthermore, we performed an experimental evaluation of the accuracy and interpretability of four algorithms using disease data. The results of our experiments demonstrate that HAC is significantly more effective in discovering clusters of regional disease.

Ontology has been regarded as the core technology of the semantic web. However, non-experts still have difficulties in participating in ontology generation. So, the growing need for public participation in ontology generation has arisen. We propose a method in which the public may participate in ontology generation by adopting the ACO (Ant Colony Optimization) algorithm. We demonstrate that the ontology generated by the proposed method is satisfactory to justify our method: precision and recall of the ontology are about 94.44 and 99.6 % respectively. The suggested method enables the construction of the semantic web environment with non-experts in the field of ontology engineering.

In this paper, we propose an effective extraction method for acquiring the interests of users from Social Network Services (SNSs). In the proposed approach, a domain ontology generated by a decision tree is first used to classify domain webpages and each user. A Social Network Analysis (SNA) method is then used to analyze the tags from the Friend-Of-A-Friend (FOAF) profiles of each user; after which, we obtained the interests of the users. The results of an experiment conducted to obtain the interests of 2012 USA presidential candidates indicate that the precision and accuracy of our approach are 91.5 and 93.1 % in classifying the users, respectively.

This paper presents a road line marker detection method in order to support night-time Blind Spot Monitor (BSM). In our BSM system, to detect vehicles in blind spot in a real-time manner, line mark detection method is necessary for the determination of exact detection window location and size. We developed a fast line marker detection method which is robust in the night-time suburb environment. Our method depends on the dividing windows and adopting local thresholds of line marker intensity for each window adaptively. We demonstrate a reliable detection performance in a real road environment and our method is light to achieve real-time operation in embedded ADAS devices.

Data visualization is a science and technology research on visual representation of the data, it puts the data into a form of visual information and knowledge, and makes full use of people natural ability of quickly identify visual patterns to understand the data. Through data visualization technology, data can be performed intuitively and clearly in various ways, and it is also convenient for data analysis researchers to find the useful information. This paper introduces the basic characteristics and the MATLAB platform graphics advantage, based on the principles of cloud image visualization, built on the MATLAB platform in the data mapping method and the visualization methods, and gives the specific Marine environment data visualization method based on MATLAB implementation, thus making the Sea Surface Temperature and Sea Surface Height and Chlorophyll Concentration these three visual products.

Data mining comes into being as a new area of research, WEB data mining technology is known as one of the major information processing technology in the future. ID3 algorithm is a often used classical algorithm in data mining technology, which is mainly applied to the implementation of data mining. It always creates the smallest tree structure and is proved that the system design has good effect to transaction analysis of log files by proofing instances, this system is effective in the log files analysis and improvement of ID3 algorithm.

The purpose of this research is to solve multiple combination and inefficient of complex dentition defect diagnosis and treatment, we put forward the thought of building data structure of AND/OR tree and set the problem node as “AND tree” and the treatments rules as “OR tree”, through introducing the positive and negative constraints to prioritization the complex rules of oral treatment and generating a diagnostic report with treatments order for medical staff make a diagnosis quickly and efficient. Results show that it is feasible to use AND/OR tree with the positive and negative constraints for reasoning data of dentition defect. Ensuring the accuracy of diagnosis in practical application and improving the efficiency of the diagnosis.

This study proposes a novel methodology to forecast traffic safety level based on weather factors by administrative district in South Korea. These administrative districts are grouped by their characteristics, such as population, number of vehicles, and length of roadways, with the use of k-means clustering. To identify major weather factors that affect traffic safety level for the clustered district groups, the random forest technique was applied. The performance of such random forest models combined with k-means clustering is evaluated using a test dataset. With the results obtained from the analysis, this study highlights that its proposed models outperform a simple random forest model without clustering.

In order to resolve the problem of unclear identification of human body region, incorrect extraction of skeleton and location of joint position, this paper mainly provides the skeleton extraction method based on depth data in self-help camera system. Firstly, we use the method of structure light to acquire the image and then use the binary way to process it to divide target from background. Then, we combine the algorithm of corrosion and expansion with the threshold method to get a skeletal structure. Finally, we get the relative position of human body, then find out the most likely bend points and locate the joint position according to the largest triangle method. The experiments results show that the skeleton extraction method based on depth data can extract skeletal structure better, and this method segments the background without unnecessary noisy point and blank to extract skeleton in an accurate way. We also design the human–computer interaction of taking photos using this method, and the system is feasible and effective.

Crashes on highways tend to be more serious on account of the high-speed driving of the vehicles involved. Therefore, a delay in handling a crash or transporting a patient may mean additional personal injury. To prevent such injury, quick emergency response is required. The purpose of this study is to analyze the optimal locations of regional emergency trauma centers, which allow for the effective handling of emergency trauma patients, and to determine the priority of each candidate hospital in terms of establishing trauma centers. Under such purpose, this study geocoded highway crash records to create a basis for spatial analysis and determined the priority of each candidate hospital by calculating the number of crashes handled based on the traffic volume and the severity of crashes using the maximum covering location problem model. The significance of this study is that it provides a basis for prioritizing the government’s financial supports for the candidate hospitals.

There have been many cases of mass mortality of fish due to the power interruptions caused by natural disasters, human errors or acts of sabotage by the employees hired by rival companies, discouraging the small and medium-sized fish farmers. Thus, in this paper, a system that can monitor the power status of fish farms to cope with such situations is being proposed. The system utilizes both PLC and ICT technologies to respond to crisis. Electric power is a critical parameter for both the land-based and the water recycling fish farms. The mass mortality of fish due to natural disasters was the highest risk in the aquaculture industry. The proposed system allows immediate response for the power interruptions and consistent power monitoring by the employees. To compensate the signal transmission loss often caused by the noise power typical in the PLC-utilized models, RUDP is proposed for the transmission layer. Also, for the Android application, a graphic user interface was designed to manage overall fish farm activities, photoperiod, seawater control, Smart Aquafarm, access control, location check, withdrawal period, and secondary battery check.

The wavelet-domain hidden Markov tree (HMT) model provides a powerful approach for image modeling and processing because it captures the key features of the wavelet coefficients of real-world data. However, it is usually assumed that the subbands at the same level are independent in the traditional HMT model. This paper proposes a modified HMT model, SHMT-S, in which a vector constructed from the coefficients at the same location of the subbands of the same level, is controlled by a hidden state. Meanwhile we also use the vector Laplace mixture distributions to fit the wavelet coefficients vector, which is peakier in the center and has heavier tails compared with Gaussian distribution. By using the HMT segmentation framework, we develop SHMT-S based segmentation methods for image textures and dynamic textures. The experimental results demonstrate the effectiveness of the proposed method.

Recently there has been various forms of sexual assault against elementary school children at various locations. In order to prevent such incidents, various educational applications for sexual violence prevention are being developed. As mobile devices are becoming more prevalent, the prevention education applications developed are optimized for the mobile devices. However, a more precise user interface design is required to enhance the efficacy of the education. This research suggests a method to enhance the user interface of an already existing sexual violence prevention application for elementary school students. It also explains a case in which the suggested user interface design method was implemented on an application according to the configuration of the proposed user interface. The results showed the learning experience was effective with the enhancements made on the user interface.

Many jurisdictions suffer from lengthy evidence processing backlogs in digital forensics investigations. This has negative consequences for the timely incorporation of digital evidence into criminal investigations, while also affecting the timelines required to bring a case to court. Modern technological advances, in particular the move towards cloud computing, have great potential in expediting the automated processing of digital evidence, thus reducing the manual workload for investigators. It also promises to provide a platform upon which more sophisticated automated techniques may be employed to improve the process further. This paper identifies some research strains from the areas of Information Retrieval and Information Extraction that have the potential to greatly help with the efficiency and effectiveness of digital forensics investigations.

The effectiveness of information security in an organization much depends on the leadership of information security manager. The leadership of information security is somewhat influenced by culture of an organization. This paper tries to explain the process how leadership of information security manager could influence the effectiveness of information security through organization culture

Context aware computing is a computing paradigm in smart space that provides quality service by being aware of users and its neighboring contexts. Uncertainty in context aware systems affects the complexity of context processing and context collection mechanism. If uncertainty exists in a context aware application, user satisfaction in the application drops sharply making it obsolete. Environment monitoring with sensor network is a good example of context aware computing. In a sensor network system, since human intervention is not possible, the level of effort to eliminate the uncertainty greatly affects the reliability of the system. In this paper, we propose a technique to cope with uncertainty in environment monitoring with sensor network. We propose strategies to identify the source of uncertainty for contexts, to represent uncertain context information and methods to process such contexts. Experiments performed on an environment monitoring system shows stable monitoring performance of the proposed technique.

Efficient compressed image data encryption is crucial issue for privacy protection because most of the images are stored in the standard compressed form. In this paper, we propose a fast and robust JPEG image encryption method. Our approach achieves efficient and robust image encryption by shuffling both spatial information and frequency domain information. Our method performs encryption in a hierarchical manner from global to local part of images. We shuffle the data while keeping the format compliance with JPEG standard because we want to apply our method on encrypting a part of the whole JPEG image. The partial image encryption is required with images having privacy-sensitive data such as the pedestrian faces or the license plates. Our proposed method uses a single 1024 bits encryption key, from which all the information shuffling and masking data are generated and applied for image encryption. The proposed method is secure against any cryptographic attack as well as perceptual information cracking. Besides, ours is faster than any of the standard data encryption method.

One of the most important works of planting crops in greenhouse is climatic factors control. In this paper, we propose to predict air temperature, relative humidity and carbon dioxide concentration in greenhouse using multiple regression models. We also propose to improve the accuracy of prediction using clustering technique. In other words, we first perform the cluster analysis and then build models for the data in each cluster. The results of predictions were compared with raw experimental data for evaluating quality of these models. The experiment results demonstrate that the proposed method outperforms the existing method. The aim of this study is to provide a feasible climatic factors control method to automatic environmental monitoring systems.

Road safety is main concern of South Korean government, as in 2010–2015, car accidents took around 25,000 people lives. It is important to understand the reasons behind these accidents in order to prevent them from happening. In this short paper, we propose to statistically measure the factors that influence on road accidents in South Korea. Specifically, we perform correlation analysis between car accident occurrences and various factors. The analysis that influences each of the presented correlation result enables government to focus their strength in the most problematic areas.

Recently, several accidents have occurred in South Korean military barracks that caused a social concern. In this paper, we argue that these accidents can be prevented. Specifically, we describe an ongoing project that applies well-known data mining techniques to predict accidents in military barracks in South Korea. For this, we first collect various soldiers’ data, such as social media, personal history and medical data, and then, use ranking, clustering, classification and text mining techniques to analyze this data.

The growing availability of information technologies has enabled law enforcement agencies to collect detailed data about various crimes. However, the volume of crime has made the process of searching and finding the useful information from the crime data difficult. In this paper, we propose an ontology-based approach for searching the knowledge in crime big data. That is, we propose to diversify the search result using ontology-based rules. In order to achieve the goal, we developed crime domain ontology and then performed the search in well-known social media such as Naver and Daum by using developed ontology.

As a famous community detection algorithm with near linear time complexity, Label Propagation Algorithm (LPA) has been an active research direction. However, that LPA regards that all the neighbor have the same importance to target node in target node’s label update process make it get stuck in poor stability. From the view of structural holes in sociology, people who fill structural holes bring social order to the network, so we improve LPA based on contact-specific constraint measuring structural holes. Experiments show that our improvement successfully detects communities with the highest stability in several commonly used real-world and synthetic networks.

An emotional design that measures human emotion for new product has been emerging in several fields. The most pervasive method of measuring the human emotion is evaluating the users’ emotions with the questionnaire including the most common emotional vocabularies. The objective of this study is evaluating the level of user’s emotion with three different questionnaires in sentence length and expression styles. In order to achieve the study objective, the main page of best award winning website from the FWA (Favourite Website Awards), an industry recognized internet award program, was selected for evaluating the human emotion. This study found that different question style has no significant effect on human emotion evaluation. The findings from this study are a lot different from the research reported in the literature outside the Korea. This implies that additional research might be necessary by adopting different target participants or investigating the characteristics of the survey respondents.

We design and implement a book recommendation system that can extract and suggest the books preferred by users through keyword-matching based on the information of frequently checked out books. The MapReduce programming model on the Hadoop platform is used to extract frequently rented books by using keyword-mapping with a target book. The MapReduce operations designed and implemented in this paper are performed to analyze the actual book-rental log data accumulated in university library, which has the characteristics of big data. An illustrative example shows that our book recommendation system can provide users with the information of the recommended books by keyword-mapping in the next book rent.

The dramatic explosion of huge number of heterogeneous medical data in smart healthcare, is leading to many difficulties on both obtaining the intelligence, cognition and natural interactions between doctors and patients. Toward to this end, this paper proposed a big medical data cognitive system with the proposed methodology that is n-ary formal concept analysis. The unique features of this cognitive system include efficient big data representation, high-quality data associations, and natural semantics interpretation among dimensions.

This review provides an introduction to MSDF in topic discovery, and discusses the status quo of the methods and applications of MSDF. This review has investigated the main thoughts of MSDF and proposed that MSDF could be divided into the fusion of data types and fusion of data relations. Furthermore, the fusion of data relations could be divided into the cross-integration of multi-mode data and matrix fusion of multi-relational data. This paper studied the methods and technological process of MSDF applicable to information analysis, especially in the competitive intelligence of scientific and technological area.

This paper describes a novel mobile motion prediction algorithm to meet the need of today’s mobile system and application which is based on Markov model. As in different time period, the things people always do usually are different, so does the route they have taken. It provides a way to constrain the path sample with time interval to enhance the prediction. In the end, the prediction accuracy is experimented up to 92 %.

Location prediction is the latest development direction in these year. This paper proposes a new method which does not need each individual history path and ID to match his/her history path with prediction path database to predict the user’s next location. In this experiment, we used two pair of coordinates to give a prediction. It’s based on the foursquare dataset. And through changing the factors that affect the location prediction, like length and time, in the general experiment, the accuracy of the prediction will be enhanced.

Testing web services is more challenging compared to traditional systems due to the absence of source code and the complexity of web services. In this paper, a dependency analysis approach is presented in web services data mutation testing. The approach at first analyzes the OWL-S (and related OWL specifications) to extract the ontology definitions associated with I/O parameters of SUT (System Under Test). Then it derives the input data syntax dependency, and semantic constraints on class, property, and parameter dependency from such ontology definitions and TCN-3 functional test suite. Base on the analysis of these dependency information, generates invalid inputs according to the dependency analysis. At last automates the test execution by using the interaction scenarios defined in TTCN-3 functional test cases to achieve invalid input injection and test verdict determination. We conducted a case study on web services use this approach. In the case study, our proposal is supported by three experiments in which web services have been tested. The preliminary results show that our approach is feasible and effective.

There are more than 12 million open source software in GitHub so far. Many studies analyzed the impact factors in development process of GitHub open source software. However, there is a lack of correlation analysis between impact factors. In this paper we propose an approach to analysis the correlation of between the speed of problem comment and the speed of problem solving. It’s proving the existence of certain factors among some relevance after experiments.

In this study, a context-sensitive grammar is suggested to analyze some patterns and configurations of RNA secondary structures. The use of context-sensitive grammar to analyze pseudoknots allows us to represent RNA structures more naturally comparing with a conventional approach of using Stochastic context-free grammar to model pseudoknots. The suggested technique directly reflects the characteristic appearance of several forms of RNA secondary structure, i.e., hairpins, internal loops, double helixes, and bulge loops.

In this study, a way of using Eclipse JUnit to perform an independent path testing and composite component testing is suggested. The suggested technique simplifies the complex arrangement of variable value setting of independent path testing so that JUnit facility of Eclipse may be conveniently used to perform a white box testing. A way of using JUnit to perform a Java component consisting of several classes with inheritance relationships is also suggested. To make a use of JUnit for the independent path testing several variables included for a testing purpose may be traced to have the similar effect of the independent path testing. Constructor arguments may be used to check right inheritance declarations to test a composite component with layered inheritance relationships. Using the suggested technique, the process to perform independent path testing and composite component testing may be simplified using Eclipse JUnit facility.