Due to the characteristic of energy constrained, it is vital to balance the energy of nodes in Wireless Sensor Network(WSN). Multipath Routing with Load Balancing(MRLB) based on Ant Colony Algorithm(ACA) is proposed. The algorithm utilizes self-adaptability and dynamic optimization capabilities of the ant colony to establish multiple paths from the source node to the destination node. It takes residual energy of the node into heuristic factor, balancing energy consumption of the nodes. A load balancing scheme is proposed to distribute the traffic over the multiple paths discovered. The scheme applies Analytical Hierarchy Process(AHP), which gives each path a certain percentage of load distribution. It makes transmission on multiple paths equilibrium, which extends lifetime of the entire network. Simulation results show the algorithm balances energy consumption of nodes and extends network lifetime effectively.
As for large-scale ontology modular partition problem, this paper analyzes the similarity between complex networks and ontology structure, converts ontology to the corresponding concept networks according to its semantic and structural characteristics. It proposes a set of quantitative criteria for measuring the local central of nodes and semantic connection strength of edges, uses these quantitative criteria to identify core concept nodes and assign weight to edges, obtains undirected weighted hierarchical networks, which explicit express ontology semantic and structural characteristics, transforms the networks to circuit voltage networks, and partitions it by the maximum potential difference method. Experimental results show that this method can acquire high-quality ontology modules in linear time and fit for the further use of ontology matching application.
Aiming at real-time decode H.264 video on IP network with packet losses, an effective error concealment algorithm is proposed depending on the analysis of high definition video streams. By use of the edge macro-blocks information of the lost slice, the motion vector of corrupted macro-block is predicted, and the error concealment is completed. Experimental results show that, compared with the error concealment of Joint Model(JM), the proposed algorithm improves the objective quality and subjective quality of reconstructed images. The algorithm does not increase the complexity of the decoding, but achieves better recovery results. It is ideal for real-time decoding for high definition video.
(α, k)-anonymity model can not thwart the homogeneity attack well because of the model ignoring the sensitive difference between sensitive attribute. It is achieved traditionally via generalization techniques. It also has some defects on efficiency and data distortion. So this paper proposes an improved (αi, k)-anonymity model. It considers the sensitive difference between sensitive attribute, and designs a (αi, k)-anonymity clustering algorithm based on greedy strategy recur to the idea of lossy join. Experimental results show that the proposed model can resist homogeneity attack and is an effective approach.
This paper proposes a new idea using the idea of the fuzzy decision. And it is based on the kernelized spatial depth and the idea of the smallest sphere, intending for the problems that kernelized spatial depth function can not have good performance on some datasets and the parameters have the influence on the effectiveness. In this way, the algorithm improves the effectiveness and robustness in outlier detection by using the advantages of the algorithm and weakening the disadvantages. This paper does some experiments on the two artificial datasets and three different UCI datasets. Results show the effectiveness of the proposed idea.
Based on the old component dependencies, this paper presents an Architecture Analysis and Design Language(AADL) system reliability model transformation method. It expands the dependencies to semantic connection, parameter connection and subcomponent calls and realizes the transformation to General Stochastic Petri Nets(GSPN). Therefore, it makes the model transformation rules from AADL reliability model to GSPN model more maturity and realizes exact and entire evaluation to reliability of embedded system.
In multi-hop Wireless Sensor Network(WSN) using clustered routing algorithm, the nearer to the sink node, the more load is put on the headers. As a result, headers near the sink node will quickly lose their energy and become unavailable which leads to network-partitioning and lower data transmission quality. In order to solve this problem, based on the analysis of existing researches, a new routing algorithm is proposed, which is based on both Unequal scaled Clusters and Redundancy of Headers(UCRH) algorithm. According to simulation results, UCRH algorithm overwhelms in prolonging network’s lifetime, reducing energy cost and supporting reliable data delivery.
Aiming at the problem of sparse user ratings facing tradition recommendation system, this paper proposes a hybrid recommendation method based on Tag and collaborative filtering(TAG-CF) to provide a solution to this problem. The neighbors set for the target item can be gained based on tagging information. It uses item-based collaborative filtering to generate the predictive ratings. By filing these predictive ratings into the sparse user-item rating matrix, it constructs a full pseudo ratings matrix. It computes the predictions based on the pseudo ratings matrix by using user-based collaborative filtering. Experimental results show that the proposed method performs significantly better than the traditional CF method.
This paper proposes a new detection algorithm based on rough set. It uses information centrality as a measure of correlation between nodes. While dealing with the boundary nodes between communities, it uses upper and lower approximations subsets so as to better simulate the real world, then it clusters nodes to certain community and identify the network to k communities, identifies the ideally community structure according to modularity, besides the k value need not to be prior given. The algorithm is tested on two network dataset named Zachary Karate Club and College Football. and experimental result shows it has high accuracy rate.
According to the data request model of power parameter estimate system, and semantic conflict type existing in the process of data integration, a heterogeneous data integration frame based on ontology is proposed to solve semantic heterogeneity problem. It improves traditional data integration structure by adding ontology semantic describing structure into data integration mediator. Based on describing the concept of domain by ontology, this structure solves the semantic heterogeneity problem existing in heterogeneous data integration, by finding semantic conflict initiative and constructing semantic mapping relations. Experimental result shows the feasibility and effectively of integration framework based on ontology proposed in this paper.
To solve the problem of traditional maximal frequent pattern mining that it can not find frequent pattern remaining more items than traditional maximal frequent pattern with the same support threshold, this paper proposes the conception of Maximal Sub-Frequent Pattern(MSFP) and relative mining algorithm MSFP-mining. The main contributions include: the conception of MSFP and analysis of MSFP character, the MSFP-mining algorithms of MSFP, such as AFP-tree, CMP-tree, SFP-tree, SFP-growth, and MSFP-tree, the superset check method of candidate MSFP and the pruning strategy of MSFP-tree, the efficiency of MSFP-tree based mining algorithms by extensive experiments. Experimental result shows that MSFP can effectively expand the scale of maximal frequent pattern.
This paper presents and realizes an IP voice system, which combines the S3C2410 processor’s low-cost, high performance features with embedded Linux software system’s high reliability advantages. It uses a unique way called four-threaded double-buffered voice data processing and delay-control playback technology to make real-time voice playback smoother, and time delay is controlled within an acceptable range. Performance analysis shows that the modular system design makes the system easier to access 3G networks by only replacing the communication module.
A new technique for the problem of incomplete data in abnormal signal detection system is proposed. Getting inspiration from the geometry, the new method compares the incomplete date with normal data, it presents a computation method of abnormal probability. With the abnormal probability, some abnormal signals can be detected directly, and the other incomplete data can be arranged. The algorithm decreases the workload and makes good use of calculation resources. Experimental result shows that when some parameters are lost, the method can get the reasonable abnormal probability of the incomplete data.
Combined with certificateless public key cryptography with proxy blind signature, a certificateless proxy blind signature scheme is proposed. The scheme enjoys the blindness properties and untraceability, and solves the escrow problem and retains the merits of blind signature without certificate. The new scheme is proved to be secure against existential forgery on adaptively chosen message attack and chosen identity attack.
Based on RSA cryptosystem, this paper proposes a new (v, t, n) fairness secret sharing scheme. In the scheme, each participant’s secret shadow is selected by the participant himself and others do not know anything about his secret shadow. Even if v(v
In the high security fields, XML documents may include information at different levels of sensitivity. It should be protected by Mandatory Access Control(MAC) policy. In order to maintain the integrality of data at high levels of sensitivity, the security labels of subjects and objects are improved. An extended MAC model called EBLP is proposed on the basis of BLP model. Security label assignment are discussed. The architecture and the access control arithmetic used to implement the fine-grained EBLP model are discussed.
This paper proposes a trust model based on Dirichlet distribution in Multi-Agent System(MAS). It uses the Dirichlet distribution to solve the limitations of a binary evaluation, the trust model can rate with graded levels. A level filtering algorithm is proposed to effectively filter a variety of malicious Agent in the referrals. Experimental results show that the proposed trust model is effective in inhibiting unfair recommendations and strategies deception.
This paper presents Cooperative Coevolutionary Estimation of Distribution Algorithm(CCEDA) to solve Resource-constrained Project Scheduling Problem(RCPSP). It integrates the cooperative co-evolutionary framework and Estimation of Distribution Algorithm(EDA), decomposes RCPSP into several sub-problems, and then applies improved EDA to cooperatively solve these sub-problems. In order to enhance the local search ability of EDA, it gives a local search method for solutions. CCEDA is compared with GAPS, GA-DBH, GA-hybrid and GA-FBI, and experimental results on PSPLIB prove that CCEDA has better performance.
By analyzing the impacts of Least Square Support Vector Machine(LS-SVM) model hyperparameter selection on the classifier, this paper proposes a method using estimation of distribution algorithms with diversity preservation named EDA-DP to optimally select model parameters of LS-SVM. Experiments are operated to recognize the benchmarks and radar High Range Resolution Profile(HRRP) datasets by using LS-SVM classifier. Compared to the grid-based method, the average recognition rate of LS-SVM classifier based on EDA-DP are increased by 4.2% and 1.76%. Experimental results demonstrate that the classifier model with EDA-DP achieves better classification ability and generalization capacity.
In order to solve the problem of conversion from the finer level of granularity to the coarser level of granularity, this paper proposes a formal method for granular synthesis based on category theory. It takes the granular structure as the granular object, the granular structure mapping as the granular morphism, then the granular object and the granular morphism composes the granular structure category. It uses the granular pushout to achieve the granular merging, and gets the granular synthesis algorithm GrSA. It illustrates the method with the concrete examples, which solves the granular synthesis issue. This method can realize the conversions between the levels of granularity.
Existing battery State of Charge(SOC) estimation methods are time consuming for the training and learning process, and it restricts the application in electrical vehicles. In order to resolve the problem, this paper uses Cerebellar Model Articulation Controller(CMAC) neural network to estimate SOC. The CMAC neural network has simpler learning algorithms and it has the ability of approximating arbitrary nonlinear functions. Experiment using the data of nickel hydride batteries demonstrate the better learning speed and convergence of CMAC method compared with Back Propagation(BP) neural network, it can meet the real time requirement in SOC, and the estimation error of the CMAC is acceptable.
In multi-core computing environment, the image processing parallel algorithms can greatly improve the processing speed. However,the existing parallel designs are focused on the specific algorithms such as edge detection and image projection, which can not form a universal design scheme. Thus, it is difficult to extend this application. Based on the in-depth study of the image algorithms parallel processing mechanism and the features of the multi-core architecture, this paper proposes an image processing parallel design scheme in multi-core computing environment, which has five steps, including analysis, modeling, mapping, debugging & performance evaluation and testing & release. The paper takes the algorithm design of parallel image Fourier transforms as an example to testify the effectiveness of this scheme in single-core, double-core, quad-core and eight-core computing environment. Experimental result shows that the proposed multi-core parallel design scheme has good scalability, and this scheme can extend the space of application for image processing.
This paper proposes an Edge-weighted Structural Similarity Index(EWSSIM), which can match well with Human Vision System(HVS). Structural Similarity Image quality assessment(SSIM) does not evaluate highly blurred and Gaussian white noise distorted images well. EWSSIM assigns different weights to contour correlation and local texture correlation of the original image and distorted image, which can represent structural similarity better than SSIM. Experimental results of LIVE image database indicate that the proposed index outperforms SSIM in blurred and Gaussian white noise distorted images and also gives a better coherent evaluation for all kinds of distortions in LIVE database.
Based on the self-similarity in vibration time series of gearboxes, local scaling analysis is developed to extract the weak information in vibration signals. Combining local scaling exponents and Principal Component Analysis(PCA), low dimensional principal components of local scaling exponents statistics are used to monitor the conditions of gearboxes. Experimental results indicate the proposed method has high efficiency and correct rate on gearbox condition monitoring.
In order to raise the level of design abstraction and simulation speed of Scalable Processor Architecture(SPARC) processor, a SPARC V8 processor Transaction Level Model(TLM) is designed in this paper. This model is based on TLM2.0 standard and interpretive instruction set simulation technology. Results of the test show that the model can execute and trace SPARC V8 programs exactly and the simulation speed of it is improved two orders of magnitude than Register Transfer Level(RTL) design.
Aiming at the actual operation requirements of CYCHU-10 cyclotron, based on Finite State Machine(FSM) theory, this paper proposes a hierarchic design method for cyclotron control system, models and simulates the system with Stateflow under Simulink environment. Simulation result indicates that it is easy to reflect the dynamic logic of a complex system with the method, and it has strong usability and reliability.
Considering the strong computing ability and a high degree of parallel architecture of Graphic Processing Unit(GPU), the paper chooses one of multiple string match algorithms based on bit-parallelism, called M-BNDM algorithm, which is to be implemented on GPU and optimized. The process for string matching is simplified to bit operation that is more suitable for data computing of Compute Unified Device Architecture (CUDA) through data preprocessing. Experimental result shows the solution is about 10 times faster than equivalent CPU algorithm. Furthermore, some factors that will infect string matching performance are analyzed.
Aiming at the test papers generating problems in examination system, a recursion random division algorithm is presented. The algorithm need input parameters for generating test papers. It uses probability density function and matrix equation to calculate parameters. It confirms the number of draw-out questions in different types and different difficulty. It uses the algorithm to generate test papers. The efficiency and adaptivity of the algorithm are demonstrated by analyzing the algorithm and evaluating the generated test papers, and the algorithm meets the requirements of examination system well.