Data replication is widely used to hide network latency, enhance response time and improve interaction experience in real-time collaboration. Users can operate on data replicas unconstrainedly, which results in consistency maintenance problems of these replicas. Aiming at the problem, this paper introduces the challenges of consistency maintenance according to the features and requirements of real-time collaboration applications. By analyzing representative key techniques of operation transformation, address space transformation and WOOT, it elaborates the challenging problems such as Undo and String transformation, and details how these key techniques are applied in the transparent adaptation from single-user applications to collaborative ones, Web 2.0 and mobile collaboration environments.
The distance between the target and the sensor has significant impact on the sensor’s sensing capacity. Recent study chooses distance as an important parameter to build sensing model, but ignores that sensor needs time to sense the target and process data. Aiming at this problem, this paper proposes an improved sensing model with temporal parameter. Compared with existing models, the model can capture the behavior of sensors more realistically meanwhile measuring the capacity of sensing moving target.
Aiming at the properties of E-commerce marketing, a knowledge-based tree-like representation is introduced to present the three factors of commodity recommendation model. The model is developed with Prolog language. It can interact data with database and individually recommend to a user based on what this user reviews in current step. The model is relatively independent. It shares the data with E-commerce system through database as interface, and can be widely adapted to different E-commerce systems.
In the light of the sensitivity of the order of data input by CABOSFV clustering algorithm, this paper puts forward a high attribute dimensional sparse clustering algorithm of the integration of sorting. The method of how to determine the two sets calculates the difference between two high dimensional sparse data sets in the first clustering, the algorithm complexity is reduced. The method improves the quality and efficiency of clustering. Simulation results of one groups of sample are given to illustrate that it can improve the quality of CABOSFV clustering.
On the hypothesis of linear system, it is a typical process of blind deconvolution for the estimation of seismic wavelet and reflectivity sequence with the observations. As sensitivity to noise for Banded Independent Component Analysis(BICA), a method called Noisy Banded Independent Component Analysis(NBICA) based on Gaussian moment is proposed, with its robustness to noise. The information between adjacent seismic traces is used for wavelet extraction. Experimental result shows that the efficiency of the method is better.
Mutual authentication mechanisms can be used in Radio Frequency Identification(RFID) systems to preserve the confidentiality of the RFID tags. In this paper, a passive attack is presented on a mutual RFID authentication protocol——ULAP(Ultra Light-weight Authentication Protocol), which is designed to resist passive attack. The attack shows that after eavesdropping a few numbers of round communication(expected number is 10), all the secret of the tag can be deduced. Therefore, ULAP can not resist passive attack.
Aiming at the shortages of beforehand estimation method, this paper proposes a comprehensive statistical method to analyze and optimize the efficiency of algorithms. Comprehensive statistical method uses multiple monitors to measure the running time of algorithms afterwards, uses Admahl law and speedup performance law to lead improvement directions. Compared with beforehand estimation method, comprehensive statistical method doesn’t need to model the hardware and software platform on which algorithms run on, and it is much more simple, efficient and accurate in practice. It can give much more detailed advices on how to optimize the algorithm, and predict the efficiency of algorithm if the optimization advices are taken.
This paper proposes a block chaotic cryptosystem based on Coupled Map Lattice(CML). The encryption order of the plaintext blocks is updated dynamically by certain rule. The updating order of sub-regions and the complex feedback association of different sub-regions make the law between plaintext and ciphertext more secret. All the chaotic maps used in the scheme are simple one dimensional. Both experimental results and comparisons with similar chaotic cryptosystems prove that the proposed cryptosystem has obvious advantages of security and high efficiency of encryption and decryption.
Probe selection plays an essential role in the procedure of designing oligonucleotide array. Considering the cost of synthesizing probes, the size of probe set becomes the most important criterion to evaluate the probe set. A good probe selection algorithm should produce a small number of probes as possible. This paper makes an improvement for the greedy algorithm of Herwig’s which is based on Shannon entropy as a quality criterion. The new algorithm is similar to the forward stepwise regression search algorithm. While adding the probes with maximal marginal effect, those probes with little marginal effect are dropped from the optimization set. Experimental results making on the data sets of HLA alleles show that the probe set produced by the algorithm is superior to the probe set selected by greedy algorithm.
In order to construct a high efficiency path finding engine for the on-line map service, a shortest path search algorithm based on spatial block path cache and index is presented. A road network resample processing is proposed to obtain a road network density image and a block algorithm based on ISODATA clustering is presented to calculate the road network spatial block region. The path cache between road network block regions is calculated to reduce the path finding node number between block regions. An index algorithm to the path cache is designed. A bi-direction path search algorithm based on the path cache and path search in block region is presented. Experimental result of on-line path search shows higher efficiency comparing with the traditional path search algorithms which can reduce the on-line path calculation time less than 0.2 s for a city level road network. It can also obtain load balance of the map application server.
In order to generate test cases set that satisfies the requirements of test, a method about generating test cases from UML sequence diagrams is put forward. By analyzing the functions use cases of system to be tested, a sequence diagram is modeled. The process of test cases generation based on this sequence diagram comes. Some constraint is added to the diagram. A scenario tree can be created by parsing it. The input, respected output, constraint and scenario context are generated through a traversing of the scenario tree, and these four also constitute the test case. The paper analyzes and summarizes the problems that how to deal with complex sequence diagrams and satisfy the sufficiency of test cases, and also brings the corresponding effective solutions.
A new algorithm based on address index called NABOAI is put forward. By using IMFI matrix, it saves the order when the item next appears in the current item which belongs to current candidate item so as to achieve the spanning type search for frequent pattern. Experimental results show that, compared with Boolean matrix algorithm, the new algorithm reduces the access of transaction greatly.
This paper proposes a multi-document summarization method based on spectral clustering. Based on clustering topic-relevant sentences in the documents together, this method creatively takes the importance of each class into consideration, along with sentence position, length and other factors to obtain the score of importance of the sentences. The sentences are sorted according to the score and extracted that meet the requirement of number of words as the summarization. Experimental results show that this method performs better than traditional methods and can improve the quality of summarization effectively.
The abstract memory model and the abstract run-time environment of executable code are described. The data flow analysis is given and the correctness and computability of data flow analysis are ensured based on abstract interpretation theory. Abstract interpretation is used to propagate the variable value range information through the generic monotone dataflow framework. New methods to automatically compute the maximal counts of iterations of the loops and to identify the infeasible paths are presented. The example of analyzing program’s data flow is given.
According to the characteristics of short message text, a clustering method of the Chinese message based on density is given. High-density region of the text data is divided into clusters and a seed queue is constructed, which is arranged in ascending order of the reachable similarity, to store the text of short message text to be expanded. The text message is disposed in a specific order. In order to make higher-density clusters to complete first, the object is selected according to a greater threshold similarity, namely that the dense space text object which can be rapidly located makes the high-density cluster complete first. Experimental result shows that this clustering method’s efficiency is increased 10 times of K-means method.
A method called AFMI based on a transaction matrix is proposed to mine the maximum frequent itemsets. The frequent itemsets are obtained from all the transactions by means of condensing iteratively the transaction matrix, the transaction vector intersections are acquired to reduce the range of search. Logical operations and pruning methods are adopted to improve the efficiency of the mining. Based on AFMI, an algorithm called AFMI+ is proposed, which can mine maximum frequent itemsets from a sliding window over data streams. AFMI+ can get the maximum frequent itemsets in current sliding window over data streams just when users need to get them periodically. Experimental result shows that AFMI and AFMI+ algorithms have better performance.
Aiming at “free-rider” phenomenon in P2P network, a new balance mechanism in which the control point is on the service providers nodes is proposed. This mechanism promises the existence of free-riders in certain degree and the inconsistency of topology is considered fully. Experimental results prove that this balance mechanism is able to use network resources effectively, and also can restraint the serious free-rider behavior.
In application of data gathering in Wireless Sensor Network(WSN), the moving Sink will break the routing at last hop, and reconstruction of the routing is a heavy spending. This paper proposes a local routing maintenance mechanism for mobile Sink. It makes use of power control to reduce the routing break and conditional refresh of the entire routing to reduce the broadcast packet, delay and energy consumption. Sink's mobile performance is increased via power control.
This paper presents a semi-fragile digital watermark algorithm which embeds the level-line, a kind of character of image, into the Discrete Wavelet Transform(DWT) domain. By considering the masking characteristics of the Human Visual System(HVS), it solves the problem between transparency and robustness. To improve the safety, it makes an Arnold transformation to the watermark before embedding. It does not need the original image when extracting the watermark, and the original watermark is not necessary when detecting and locating tampered areas. Experimental results show that the algorithm has good invisibility and high robustness against non-malicious attacks, and it can locate the tampered areas roughly.
This paper proposes an improved Aho-Corasick_Boyer-Moore-Horspool(AC_BMH) algorithm, which utilizes double-character skip for both larger pattern strings mismatching possibility and further jumping distance, and combines Quick Search(QS) algorithm for even longer jumping distance when pattern strings matching fails. Compact storage mechanism is employed to decrease the amount of memory usage. Experimental results show that the matching speed of string is improved about 29%~52% with the improved algorithm, and the amount of memory used reduces about 90% when many pattern strings exist.
This paper analyzes Windows 7 secure boot technology. It researches the boot flow of BitLocker drive encryption based on trusted computing. Aiming at the BitLocker and Bootmgr secure vulnerabilities in the implementation process, BitLocker and Bootmgr circumvention schemes and Address Space Layout Randomization(ASLR) coping scheme are designed in order to prove that the secure boot measure of Windows 7 can still be broken through by malicious software such as Bootkit.
This paper proposes a novel Deterministic Packet Marking(DPM) scheme to trace the cross-regional source of Distributed Denial of Service(DDoS) attack. It adopts a packet marking algorithm inspired by CRT, and combines DPM-RD scheme to achieve better performance. Theoretical analysis and simulations reveal that the increasing number of attacks do not effect the false positive rate significantly, and compared with similar schemes, the scheme can reconstruct attack path in a comparatively short time with less packets.
Feature extraction of structure irregularity for skin lesion boundaries has a great significance in computer-aided diagnosis for melanomas. Based on previous work using local Fractal Dimension(local FD) for contour irregularity descriptions, this paper proposes the feature extraction method of boundary irregularity using multi-scaled curvature analysis. Feature differences from adjacent scales are used to enhance the discrimination power between malignant and benign shin tumors. Comparative experiments of both local FD and curvature analysis show that the features difference of adjacent scales has characteristics of larger inter-class Hausdorff distance with the scale increase, while local FD performs better than curvature analysis with larger Hausdorff distance value. Experimental results show that the two methods not only have strong classification capacity of structure irregularity, but also are helpful to weaken texture irregularity for the classification.
This paper integrates the Boosted Cascad algorithm with the Histograms of Oriented Gradients(HOG) features to achieve a fast and accurate pedestrain detection system. Through using the structure of the Boosted Cascad algorithm, the areas of information less are removed, the amount of computation are reduced. Experimental result shows that the pedestrian detection efficiency of the new method is high and has strong robustness.
An Adaptive Media Playout(AMP) algorithm based on Proportion Integration Differentiation(PID) control(AMP-PID) is proposed to reduce the video playout jitter caused by buffer underflow. It combines the control theory of incremental digital PID algorithm and the adaptive media playout algorithm, and adjusts the duration of each frame according to the channel state and the data cache size. Also, it controls the play rate range and the play rate variation in reducing the probability of buffer underflow at the same time to achieve a smooth video playout. Simulation results show the algorithm has better performance than AMP-Mean algorithm in reducing the probability of buffer underflow, achieving a smooth video playout and reducing the playout latency.
This paper fuses the features of both color and edge in particle filter tracking framework, and proposes an improved tracking algorithm by introducing several optimization factors, such as describing the color feature of target using kernel density estimation under HSV color model, using more practical motion model in particle filtering, clustering the particles with mean shift algorithm, and reordering the steps of clustering and resampling. Additionally, it also improves the edge matching method by introducing mean shift. Such method adds a pretreatment of the edge information into the matching process that makes the distribution of particle weights more rational. Experimental results show this algorithm has better robustness and real-time ability.
Slow speed and low accuracy of detection are the common drawbacks of traditional video moving target detection methods. In order to overcome these shortcomings, a new detection algorithm of moving targets is presented which integrates frame difference with variational level set method. The motion region is initialized through this improved frame difference method, and then it is used as initial level to evolve by the level set equation without re-initialization. In the evolution process, image intensity and optical flow information are used to control level set curve stop at the object boundary. Experimental results show that the proposed algorithm is an effective video rigid moving object detection method with high detection speed and high accuracy.
This paper proposes a novel architecture of convolver which is based on RAM. It is power and area efficient compared to the conventional FIR architecture which is based on registers. The reduction percentage of power and area increases with the order of the convolver. It implements two pairs of convolver with different orders. Each contains a RAM based one and a FIR one for comparison. Simulation results by DC and Prime Power show that for the first pair with order 63, area and power reduction percentages are 10.1% and 8.4% by employing RAM based architecture, while for the second pair the above reduction percentages are 14.9% and 15.2%. The chip test result when the first pair tape out successfully shows that the power reduction percentage is 7.9%.
This paper presents a new NoC architecture. By adopting the virtual output queue technology, this architecture completely resolves head of line blocking and deadlock which exists in traditional ring topology architecture. By changing ring number, flit sending speed and Buff depth, it evaluates the performance of this architecture and compares with 2D Mesh in the experiment. When node number is 16, the performance of new bidirectional ring topology NoC architecture whose Buff depth is 2~4 flits is better.
Aiming at the common problems including difficulties in product content analysis, low density in customer scores and new customer recommendation, existing in recommendation technologies today, this paper designs a blending recommendation technology which employs demography analysis technology based on cooperating filtering technology, and implements a recommendation model instance in a Web resource system. Experimental results indicate that this technology can solve the problems mentioned before, and efficiently improve recommendation quality comparing to the traditional technologies.
To improve the safety of real-time system at transient overload, a schedule algorithm based on mixed priority is proposed. The priority of the algorithm consists of relative deadline priority and relative slack priority. With relative slack, whether tasks will miss deadline can be predicted. Through the complete CPU preemption, frequent context switch can be avoided. Simulation results show that the presented algorithm improves the efficiency of task schedule and effectively decreases the Missed Deadline Percentage(MDP) at transient overload.
According to the knowledge of product configuration, this paper proposes a model of mass customized product configuration , which is based on ontology and includes both configuration and constraint types. It defines every concept and its relation of the model, describes them formally in Web Ontology Language(OWL), expresses specific constraint rules in Semantic Web Rule Language(SWRL), and checks the efficiency of the model with Protégé.
This paper presents a query model based on semantic meta-model to search digital resources. By improving Top-N algorithm in traditional relational database, a novel similarity evaluation method is proposed based on the different kinds of data type and the semantic meta-model. Based on the method, an intelligent querying system is developed, which is used in Shanghai Educational Resources Center. Actual application shows that it can improve the query precision effectively.
Location methods based on Received Signal Strength Indicator(RSSI) are adapt to RFID systems because they are easy to realize and have low overhead. This paper presents implementation method and applications of RSSI in RFID reader of UHF band. By using appropriate measurement means and analyzing the radio propagation fading model and a large number of measured data, it proves that the RSSI of tags can work on real-time location for passive tags.