Abstract:The frequency of CPU has boosted in recent years to meet the requirement of modern application on compute performance. However, the "Moore Law" can't be maintained only by increasing the frequency of single chip, the homogeneous multi-core system has appeared to compensate this insufficient. With the support of homogenous multi-core processors, the chip puts multiple processors with same status and structure together to minimize the cost of performance improving, load balancing and fault tolerance. The performance of parallel system is doubling when combining with appropriate software. In this paper, our research is focused on the multi-core task scheduling along with the changes in processor architecture from operating system level. The system uses hybrid scheduling which is composited of independent inter-cluster scheduling and unified inner-cluster scheduling. We do some deep analyzing on multi-core scheduling theory and implementation strategy from different aspects, including scheduling model, scheduling algorithm, dispatch algorithm and scheduling occasion. The experiment results prove the correctness and schedulability of the algorithms.

Abstract:RCC cache coherence protocol in Godson-T many-core processor is a characteristic parameterized concurrent system. It is a challenge to verify this protocol. Cubicle is a recently built parameterized model checking tool based on SMT solver. We used Cubicle to model and verify RCC protocol successfully. The experimental results show that RCC protocol satisfies all kinds of safety properties regardless of how many nodes it has.

Abstract:Software cost estimation has been an important part of the software project management. Afterover half a century of research and industrial practice, cost estimation methods and models have been greatly enriched. A variety of cost estimation tools are derived from these methods and models. However, cost estimation methods and models are based on historical project data. Companies and organizations without historical project data can only use data from other companies or organizations to do the cost estimates of their own projects. How to effectively do cost estimationwith cross-company data becomes more realistic. Aiming at this problem, we propose a new cost estimation method to use cross-company data effectively. The experiments show that the method could improve the accuracy.

Abstract:With the development of the Internet and e-commerce, users can write their opinions about the product on the website after they buy it or use it. There are a large number of product reviews contain abundant information which can help manufacturers and users to make important decisions. Based on semantic and linguistic analysis, this paper proposes a method which can extract product features from the online reviews and rank them based on customer attention strength. At the same time, the sentiment orientation users have on the feature is determined according to the polarity scores of opinion words. The online reviews about notebook Computers are used as experimental data. Experimental results show that the method achieved good precision and recall rate.

Abstract:Dynamic voltage and frequency scaling technique (DVFS) has been introduced to manage system power at software level. In this paper, we firstly discussed the disadvantages of commercial CPU frequency governors on interactive systems. Then, we proposed and implemented a new DVFS scheme aimed at Interactive Systems. The scheme is designed to save system power based on the premise that user experience must not be declined. Experimental results demonstrated that most application can save more than ten percent power compared with the original Ondemand DVFS governor, and even more than thirty percent for some other applications.

Abstract:The conflict between stability and growth has existed in companies for long. In contrast to traditional compromising solutions, Throughput Accounting (TA) offers a win-win way that meets both requirements from a systemic view. This article introduces the concept of TA and proposes a throughput-viewed metric hierarchy for supply chain management, and a relevant decision-making system is designed and implemented with Business Intelligence Technologies. The feasibility and validity of the system has been verified practically.

Abstract:Users' individual requirement has been increased with the popularity of intelligent terminal. In many scenes, users need to collect and consider much information when they make a choice, then achieve this goal by accessing web services. However, the problem for collecting information is that there are amounts of information, and the information have hidden relations, these factors make users hard to make decisions. In addition, when users search services they have to face the fact that single service always can't satisfy their needs, they need to compose many services to achieve their target. The above two questions had bad effect on users' demands satisfaction, therefore the user-centric intelligent service composition become more important. To solve this problem, we propose a user-centric portal system for mobile information services, it focus on intelligent service composition, and it can help users to find specific target and compose services to achieve the target by using individual information and context data.

Abstract:There are mechanical, electronic or network vehicle anti-theft systems. But in reality its flexibility and the interaction between the drivers and vehicles are poor. In this paper we designed both the hardware structure and software on Android OS to achieve the theft proof function. The stolen car's owner can implement human-car interaction, receive the car image, locate his car automatically and plan the optimal chasing path. The experimental results show that the system has good flexibility, strong interactivity, small size and low cost, which meets the demands of vehicle anti-theft system.

Abstract:Based on web pages full-text search technology. Design a full-text retrieval system to meet the special demand of e-textbook. It contains teaching materials document processing module, the indexing service module and the retrieval service module. According to the special feature of e-textbooks, we define the data structure of the index file, data structure of the text file, data structures of index entries and the score model for ranking. The retrieval system provides a fast and accurate search results to the learner.

Abstract:Fault localization is extremely important for fault tolerance in Virtualization-based cloud platform. This paper proposes an inter-node dependency-based log analysis approach consider the time-varying spatial topological using decision tree. Dependencies between nodes can help compressing log, providing valid information for analysing, thereby completing the fault localization faster and more accurately. Experiment results show that our approach can help to achieve effective and efficient fault localization in the larger-scale virtual environment variable.

Abstract:A microcontroller-based greenhouse illumination control system is designed in this paper. The system utilizes the digital light sensor (BH1750FVI) to detect the illumination inside and outside. According to appropriate light range of the different crops, the system can select to block the shading film or open a certain number of complementary lights. This system has automatic control and manual control modes. The automatic control mode can automatically maintain illumination in the desired range according to the selected type of crops in greenhouse. In manual control mode, shading film and light needs to adjust manually. Tests showed that the system can compensate for manual control error and low efficiency defects and has strong practicality.

Abstract:With the rapid development of Internet technology, we have now entered a service-based Internet time. Location services is a widely welcomed and development service, but it has some problems—high complexity and weak Performance. To alleviate these drawbacks, this paper proposes a solution based on REST. After testing, it can use simple URL to get resource. And it can also pass 100 clients 10000 requests test.

Abstract:According to the problem of insufficient performance ability for water pollution transport simulation, by adopting to the digital earth technology and complexity theory, the data resources of water pollution event was integrated. The trunk stream of watercourse was decomposed into multi-scale resolution. A water quality complex agent was established with characteristics and behavior patterns. And the efficiency of the migration of calculus was improved by introducing of high-performance computer technology. Then a water pollution transport simulation framework was constructed based-on the digital earth technology which merging together the water quality model and having high expressive force. Results show that this framework has enhanced the ability of spatial representation of pollutant transport process, and provided an integrated platform for pollutant transport simulation and scenario analysis.

Abstract:According to the problem of informatization construction of domestic aviation logistics, a suit of aviation logistics query system based on the IoT is presented and designed. The data was collected by using the technology of RFID and transmitted to ZigBee terminal node by RS232 serial port. Then the data was transmitted over short distance to routing node by ZigBee wireless network, and transferred out through a ZigBee gateway at last. Experimental results show that the platform can manage air cargo intelligently, visually and informationalized. The design provides new idea and direction for aviation logistics management and has a good prospect.

Abstract:As the key problem in cloud storage, security is always drawing the users' attention. A safe and efficient cloud storage scheme is designed for users' identity and data privacy protection in cloud storage security. In the scheme, a storage structure based on time series is constructed and designs a key-deriving algorithm based on the structure. It can not only be good for the generation and management of the key but also can save the storage space. By analyzing the security of scheme, it indicates the scheme can insure the security of data and can protect the privacy of users' identity.

Abstract:RSKNN is an improved kNN algorithm based on variable parameter rough set model. The algorithm guarantees under the premise of a certain classification accuracy, effectively reduces the computation burden of the classified samples, and improves the computation efficiency and precision of classification. But in this algorithm,the instances of each class are simply classified into core and boundary areas. It has the limitation that it isn't classified according the features of datasets. An efficient algorithm aiming at learning multi-representatives for RSKNN is proposed. Using the theory of structural risk minimization, a few factors that determine the expected risk of new classification model are analyzed. And an unsupervised algorithm for partial clustering is used to build an optimal set of representatives. Experimental results on UCI public datasets demonstrate that the proposed method significantly improves the accuracy of the classification.

Abstract:Bug location is the key step in the software defect repair. With the rapid development of complex computer software and network, how to fast and efficiently locate bugs related code has become a pressing problem. This paper studies the existing bug localization method based on information retrieval technologies and a two-phase bug localization method based on defect repair history was proposed. The method pays more attention to the defect repair's local phenomenon. This method does not only rely on text similarity, but also more concern about the defect fix behavior, change of information, code characteristics and other factors. Based on information retrieval and defect prediction this method is proposed to improve the accuracy of defect location. At the end of the paper, two open source projects are conducted to validate the feasibility and effectiveness of the method.

Abstract:Based on ViBe, this paper explores a new moving object detection algorithm combining with the three frame difference method. Firstly, we build model on every pixels of the background with the advantage of ViBe; then, do the logic AND operation between two differential images which have been subtracted from the current image and the previous image; lastly, update the model in real time with the idea of ViBe. Meanwhile, in order to remove the high frequency component of the image, we add wavelet denoising to every frame of the image. This algorithm effectively overcomes the effect of illumination change on the system and eliminates the ghost as well as the blinking background pixels. Experiments confirm that this algorithm can accurately extract moving objects in multiple environments and has higher robustness.

Abstract:The locating method of DPM 2D barcode image is studies in this paper. Because of the issue of reflective light of metal surface and the uneven surface, the DPM 2D barcode image collected will be of uneven illumination, low contrast, high light and other issues. These problems will impact on locating of DPM barcode greatly. In this paper, the homomorphic filtering method is used to enhance the image and the gradient projection method is used to locate the barcode region in preliminary. Because of the similar gray level between background and barcode which will result in segmentation error in the traditional thresholding method, an adaptive binarization method composed of Otsu method and neighborhood threshold method is used in this paper. Then the mathematical morphology method is used to fitting the "L" shape of Data Matrix barcode. Finally the convex hull algorithm is used to locate barcode region precisely. Experiment result shows that this method can solve uneven illumination and low contrast effectively and locate the barcode region quickly.

Abstract:The goal of scene understanding is to recognize what objects in the image and where the objects are located. Hierarchical structure is commonly found in the natural scene images. This structure not only can help us to identity the objects but also how the small units interact to form the whole objects. Our algorithm is based on the level structure. We merge the neighboring segments continuously until they combined into the whole object. The result is a forest which contains several trees, one tree commonly represents one object. We introduce a machine learning model to describe the merge process, greedy inference to compute the best merge trees, and the max margin to learn the parameters. We cluster the segments features to initialize the parameter. The experiment result could be accepted.

Abstract:Recently, there are some problems like extracting hardly and lacking classification methods in stylistic classification of social networks. Combining network stylistic diversity, multi-attribution and dynamic characteristics.A attribute fusion and thesaurus associated method based multi-agent has been proposed from feature extraction. Firstly, it extracts the basic attributes of keywords and meaning of characteristics. Then, a multi-agent fusion classification model has been established with the interaction of multi-agent and it also gives the algorithm of the model. The experimental results show that this method which compares with the traditional single fusion classification classifier and other multi-classifier fusion classification not only achieves the high-precision network stylistic classification in semantic network through Semantic features extraction but also receives Social Network stylistic classification's automation. The method has a higher accuracy classification and stability.

Abstract:or some reasons such as dynamic network topology and limited energy of nodes, the multicast routing algorithm of Ad Hoc network must be aware of the energy status of network nodes and be very self-adaptive to alter the multicast route as soon as possible. In this paper, the ant-agent of AntNet algorithm can gather the energy status of paths and update the pheromone. As a result, the multicast route can evaluate the changes of energy in whole network. The results of simulation show that this algorithm can balance node energy and improve the performance.

Abstract:Traditional person identification methods have a low tolerance for occluded situation and complicated background. We focus on an image from an overhead camera.we utilize depth information for the person identification. We apply three features to the identification method: (1)the height, (2)shoulder width, (3)depth histogram. Experimental result shows our method has higher accuracy, strong robust-ness and good real-time with the existing methods of person identification.

Abstract:With the development of Internet technology, the Intelligent Answering also has drown more attention, because it can solve user unsure questions timely. Intelligent Answering system typically includes comprehension questions, information retrieval and answer extraction and selection. Sentence Similarity Computation is a part of comprehension questions, which directly affects the accuracy of the final answers. This is based on the morphological pattern and improved edit-distance method, adding semantic information of words and taking weight into account to the new sentence similarity method in. And Intelligent Answering is implemented, to make the correct rate of the system greatly improved.

Abstract:In order to solve the case of vehicle video background changes in real time vehicle detection and tracking issues, this article raises a rear axle feature-based vehicle identification and tracking algorithm. Its characteristic is using innovative rear axle feature. Axial characteristics are specific for the two in the rear brake light for the endpoint, license plate at the end line of the axis of symmetry. Algorithm will calculate its histogram information after obtaining the target area and particle filter tracking. Finally, we test the feasibility of algorithms.

Abstract:Shapelet provides a fast classification method in time series classification, but the extraction of time series Shapelet is so slow that it restricts the application of the Shapelet. In order to speed up the extraction of time series Shapelet, an improved method is proposed based on the principal component analysis. Firstly, it uses the principal component analysis (PCA) to reduce the dimension of time series data set and chooses the reduced data to represent the original data. Secondly, it can extract the most discriminatory Shapelet sequence from the reduced data. Lastly, the experimental results show that the improved method improves the speed of the extraction and ensures the accuracy of classification.

Abstract:Based on the observed data of chamber earth pressure from on-site construction of earth pressure balance shield, a time series of chamber earth pressure is created and handled by singular spectrum analysis method, including singular spectrum decomposition and reconstruction, and a subsequent forecasting of the time series is performed based on the linear recurrence relations model. It is turned out that singular spectrum analysis can effectively separate principal components from time series of chamber earth pressure, especially in the extraction of weak periodic components. In addition, it is available for singular spectrum analysis method to reduce noise jamming by filtering out a part of the original time series that is supposed to be noise rather than signal. Consequently, it is reasonable to make the forecasting more accurate and smooth by applying singular spectrum analysis method to the analysis of chamber earth pressure values, which would be helpful to the future predictive control of chamber earth pressure in earth pressure balance shield.

Abstract:In view of the present electric curtain control system for the lack of control accuracy, instability of the system error, a fuzzy self-tuning PID control algorithm was proposed and used in the system control. The system took CC2530 as the core of communications and control, which was embedded in the fuzzy PID control algorithm. The users' requirements could be responded quickly and accurately. Eventually it could control the curtain under the users' command and complete the manually smart control to the desired degree. After a system simulation test, the output of the system was not up to the overshoot and it acquired the setting time of 0.18s and steady-state error of about 0.15%, which were less than the conventional PID control. Thus it has certain practical and theoretical research reference value.

Abstract:Uncertain data exists in many situations, such as sensor networks, mobile computing, military, telecommunications and other applications, which makes it difficult to apply traditional algorithms to mining frequent item sets. To deal with these situations, we propose an efficient algorithm based on possible world model with single scan of database. The algorithm works well without any tree construction. Experimental results show that the efficiency of our algorithm is better than UF_Growth.

Abstract:In the massive data retrieval applications, hashing-based approximate nearest(ANN) search has become popular due to its computational and memory efficiency for online search. Semi-supervised hashing (SSH) framework that minimizes empirical error over the labeled set and an information theoretic regularizer over both labeled and unlabeled sets. But the training of hashing function of this framework is so slow due to the large-scale complex training process. HAMA is a Hadoop top-level parallel framework based on Bulk Synchronous Parallel mode (BSP). In this paper, we analyze calculation of adjusted covariance matrix in the training process of SSH, split it into two parts: unsupervised data variance part and supervised pairwise labeled data part, and explore its parallelization. And experiments show the performance and scalability over general commercial hardware and network environment.

Abstract:Failing to adapt to dynamic changes and depart fromlocal optima are two disadvantages of basic group search optimizer (GSO) in the dynamic environment. A sensitive individuals and multi-producer based dynamic GSO named SMGSO is proposed in this paper for dynamic optimization problems. Firstly, sensitive individuals are introduced in GSO in addition to producer, scroungers and rangers, which are responsible for detecting the environmental change. If environmental changes are detected, some individuals are initialized to respond to them. Secondly, a new update model of scroungers is proposed based on the center of multi-producer to improve local search ability. At last, arole assignment strategy based on population diversitywhichis beneficial for keep stable diversity is adopted to determine the ratio of scroungers to rangers. Experimental results demonstrate that SMGSO is superior to other heuristic algorithms in dynamic environment, which may not only find the optima as possibleas closely but also trackthe changed optimatimely.

Abstract:An image restoration based on Radon transform and minimizing variational function was proposed, inorder to restore the blurring and noisy images by photography. The proposed algorithm firstly de-noised the blurring and noisy image by directional filter, then estimated the spread point function based on Radon transform which can eliminate error of the estimation due to the directional filter. Since then, the original image was de-burred by de-convolution with the spread point function. Finally, the de-blurred image was further de-noised by the variational function with L0-norm regularization. By experiments, the proposed algorithm can restore blurring and noisy images.

Abstract:The airborne active self-protect jamming becomes one of the most effective defense measures with the rapid development of EW (Electronic Warfare). The detection effectiveness and survivability of radars in the battle is seriously threatened. According to the demands for assessing the detection effectiveness of radar, detection probability is selected as a quantitative evaluation index. Firstly, via the organic combination of different modules resolved from the detection process of radar, the detection model is established and the mathematical models are recommended in detail. Then functional simulation of the radar detection process is devised, and the system based on HLA is realized. Lastly, detection probability with its target distance is obtained online through the simulation, and it is used for assessing the detection effectiveness of radar quantitatively. The result indicates that this model and method is feasible and effective, and it could provide some reference material for the research to radar counter-counter measures.

Abstract:Speech exists additive noise which is the main reason for reduced MFCC robustness. Then reduced MFCC robustness makes the speaker verification system identification rate decreased. To solve the problem, multitaper technology is introduced, which is a little effective in improving the noise robustness of MFCC. To get a better result, we propose a improved multitaper MFCC extraction algorithm which combines multitaper and spectral subtraction methods. Multitaper technology improve the robustness of MFCC, then spectral subtraction method enhance the speech and reduce the effect of noise. Simulation results show that when the test speech contains the additive noise, new algorithm could achieve better result in EER and minDCF which are the evaluation index of speaker verification system.

Abstract:OAuth is an protocol used to identify the client, and control resource access. Because it concerns about the privacy of the private resource owner, the security of OAuth protocol is fairly important. This paper mainly research on the OAuth2.0 protocol text, and make an abstraction on it, build an model in AVISPA, an formal verification tool for security protocols, and then verify the model in AVISPA. Finally, we find there is an attack mode that may result in leaking the private resource to attackers. We suggest a way to model the message to be authenticated as a symmetric key, which is innovative. This modelling and verification method we used on analyzing OAuth2.0 can be used in the verification of other security protocols, like the online payment protocol.

Abstract:Software vulnerability is one of the important causes of computer security. Taking the CVE-2012-0158 as an example, the form prince and exploitation way of vulnerability is explored. Which codes and functions are called by the procedure when the vulnerability is triggered are briefly described through the dynamic analysis method and the causes and hazards of vulnerability are explained to arouse people's attention of taking safe development and avoiding vulnerability. Then safe development methods based on security are mentioned to reduce computer security problems caused by software vulnerabilities fundamentally, so as to improve the safety performance of the system and software.

Abstract:Through the multiple-backup strategy HDFS can restore data easily when data is damaged or missed. However, the data stored in system increases all the time. When the data scale has become very big, the strategy will need several times of storage space to store the backup data. This article proposes to use erasure codes to replace the multiple-backup strategy, which can greatly improve the storage efficiency and reduce extra storage expend.

Abstract:As a fundamental industry, petroleum industry plays a very important role in the development of the national economy. Several major oil companies, the large and medium-sized state-owned enterprises, generate millions of daily business documents,and millions of workflow to initiate and approve. How to manage of these large enterprise content efficiently and safety has become an urgent issue. Introduction of the concept of enterprise content management is no doubt one of the most effective ways to solve this problem.FileNet is an important part of IBM enterprise content management solutions. It provides enterprise-class scalability and flexibility, and can be seamlessly integrated with content management and work flow management. It provides a leading global integrated environment for business documents and work flow management for customers. With analysis on the content management and work flow management, the characteristics and powerful functions of FileNet in this article, we elaborate how to achieve efficient management among massive business documents and work flow of petroleum enterprise through the FileNet enterprise content management solutions.

Abstract:Understanding and implementation of automatic-fire-alarm-system's output communication protocol can impact on two sides' communication compatibility. This paper introduces the protocol's environment and functional requirements, analyzes it using formal analysis method, gives its Petri net model, states the requirements implied in the protocol and affecting compatibility. It makes some suggestions on using the protocol for networked applications.

Abstract:Uygur is analyzed from the angles of the library, computer, database, solve the uygur language in a large database of ORACLE storage. This paper solves the displaying and the retrieval problems of the Internet, with national language retrieval machine readable catalog, from the aspects of interface, tips, system information such as national language localization.

Abstract:Biodiesel and regular diesel in any proportion miscrible mixtures can be used instead of petrochemical diesel fuel for internal combustion engine. In order to make the mixture in internal combustion engine applications to achieve good effect on performance and emissions, we need to research different proportions of biodiesel-diesel mixture atomization effect. In this paper, digital image processing methods is used for biodiesel-diesel mixture atomization image processing, to obtain the number per unit area and the average diameter of atomization particle, fast processing speed, accuracy, reducing the traditional manual method of analysis of subjectivity and a higher failure rate.

Abstract:License plate recognition is an important branch of the intelligent transportation system. In the character recognition module, there is a misrecognition problem. The reasons are the methods of the recognition and the singled template. To solve these problems, a improved LBP feature operator is proposed, extracting features from the segmented characters, thus improving recognition rate. On the other hand, inclining or deflecting templates, to adapt to different angles of inclination of the plate image. The experimental results show that the recognition rate has been improved by our methods.

Abstract:This paper gives a model about satellite application information chain based on queuing network, and ensure its basic parameters. It analyses the question about satellite application information chain's point resource disposing based on queuing network, and optimal algorithms based on queuing time are presented, to dispose the point resource.

Abstract:In order to solve the problem of the communication between different systems on the network, we put forward the model of the network connection. Aiming at the default value in network connection model that cannot be directly obtained by the HTML parser from the web page, this paper presents a solution based on ontology reasoning, which includes analyzing the web page with ontology, building ontology model for the page, defining rules in KAON2 ontology reasoning machine, and reasoning out the default value of network connection model. Experiments take an online judge system as an example. The results show that the method has high recognition rate and greatly improves the automation degree of the network connection model.

Abstract:During data recovery, it would cost a lot of time to scan from begin to end, and most of the recovered data would be not required. Restoring the particular data files can be useful to save time and improve efficiency. Paper begins with presenting the principle of file reading and writing, and analyzes the OS' actions when the file is modified or deleted, as well as the changes of the File Associate Table (FAT) and File Directory Table (FDT). Aiming at the problem that the current data recovery method is not in effect for the lost files without continuous clusters, we can do the job for the particular file according to its clusters' distribution characteristics, and paper tests it in some experiment.

Abstract:Aimed at the problem that the present precise positioning technology of coal mine is not mature, a real-time positioning system of locomotives based on acceleration sensor is presented. On the foundation of locomotives of its information and scheduling information combined, the system introduced the positioning principle and concrete implementation scheme. It also analyzed the three key techniques for realizing this system: the acceleration sensor calculating the shift accurately, judging the driving direction accurately, the low bit error rate of wireless transmission. And it solved the key technique problems through designing and improving the correlation algorithms. In addition, the feasibility of the system is proved through simulation and testing experiments.

Abstract:This paper focuses on the effects of light condition, glass-wearing on driver's eye, and proposes a way of human eye detection by Hough transform and neural network classifier. Firstly, two eye candidate regions were selected based on the geometry and symmetry of the iris. Then, coarse human eye positioning is conducted by edge detection and MAE operator. At last, the B-P neural network was utilized to pinpoint the human eye. For six groups of video images captured in three different situations, that is, different light, different backgrounds and different skin color, three groups were performed simulation experiment using matlab7.0. Results show that the algorithm for complex situations human eye detection has a strong robustness, improve the accuracy of eye detection greatly.

Abstract:In the actual life cycle of software, due to various reasons, the time that spending on software testing is often not enough to perform the comprehensive test. Coverage ratio and the number of test cases as the main quality factors, have been payed more attention to. However, the execution order of test case as one of the important factors that affect test efficiency, has not caused enough attention. This paper presents a test case generation technology based on genetic algorithm, which considering Coverage ratio, the number of test cases and the execution order. Testing personnel can execute test case in the established order and accomplish the most important software function test in the time as shorter as possible according to the method referred in this paper.