Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2022-05-10 Cooperative journals: 《计算机应用研究》
Abstract： Abstarct: To address the problem that a new type of DDoS attack, link-flooding attack (LFA) , is difficult to detect, an LFA detection method based on MS-KNN (Mean Shift-K-NearestNeighbor) method in SDN is proposed. Firstly, this paper simulated LFA and constructed LFA dataset by building an SDN experiment platform; secondly, an improved weighted Euclidean distance mean shift (MS) algorithm was used to classify the LFA dataset; finally, the K-nearestneighbor (KNN) algorithm was used to determine whether LFA data were included in the classification results. The experimental results show that the use of MS-KNN not only obtains a higher accuracy rate but also a lower false positive rate compared with the KNN algorithm.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-01-03 Cooperative journals: 《计算机应用研究》
Abstract： The research of user identification across social networks not only provides a basis for multi-network data fusion, but also has a wide range of applications in user identity monitoring, rumor control and other fields. Aiming at the problem of ignoring the role of user interest in user identification across social networks and the high time complexity, this paper proposed a user identity algorithm based on user interest (UI-UI) . Firstly, the proposed algorithm filtered the user nodes by Blocking to improve the efficiency of the algorithm and reduce the time complexity. Secondly, it modeled the user's interest according to the user generated content(UGC) and user social relations, and used the similarity of user interest as the basis for user identification. Finally, it used the method of semi-supervised learning for user identification. Experiments on real social networks show that UI-UI algorithm can effectively identify cross-network users, and both the accuracy and recall rate of the algorithm are stable, besides, the running time is significantly reduced.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-20 Cooperative journals: 《计算机应用研究》
Abstract： For implementing the synchronization optimization of tasks execution efficiency and execution cost, a multi-objective scheduling optimization algorithm of DAG tasks in cloud environment is presented. Our algorithm defines the multi-objective optimization problem as the trade-off optimal solutions set satisfying Pareto optimal and solves this model by the heuristic method. At the same time, for evaluating the quality of multi-objective trade-off solutions, a evaluation mechanism based on hypervolume method is designed, which can obtain the trade-off scheduling solutions with conflict objectives. Through setting cloud environment and three kinds of synthetic workflow and two kinds of real-world scientific workflow, we construct some simulation experiments. The results show that, compared with the same type of single objective algorithm and multi-objective heuristic algorithm, our algorithm not only has higher solving quality, but has better trade-off degree of solutions, which can conform to the mode of resource utility and workflow scheduling in real-world cloud.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-20 Cooperative journals: 《计算机应用研究》
Abstract： In wearable fall detection of the elderly, too much characteristics will cause the curse of dimensionality, and affect the accuracy of subsequent fall detection. To solve this problem, this paper uses time domain analysis method to extract feature vector. Then the proposed improved kernel principal component analysis (IKPCA) algorithm is used to reduce the feature vectors, so as to obtain high-quality feature vectors, which makes the subsequent classification more effective. IKPCA algorithm firstly uses the I-RELIEF algorithm to select the initial feature vectors, then calculate the information measure and similarity measure of the falling feature vectors. Finally, according to the similarity measurement of the falling feature vectors, the invalid falling feature vectors are eliminated. The IKPCA algorithm not only keeps better dimensionality reduction ability of the Kernel Principle Component Analysis (KPCA) algorithm, but also expands better classification ability. It experiments on real data sets. The comparative analysis shows that, compared with other algorithms, the IKPCA algorithm can obtain higher-quality feature vector data set.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-18 Cooperative journals: 《计算机应用研究》
Abstract： This paper studied the network connectivity modeling for vehicle delay tolerant networking. Firstly, it assumed that the process of vehicle entry into the road obeyed the Poisson distribution and the speeds of the vehicles on the road were the Normal distribution. And then, it derived the time interval distribution between two adjacent vehicles based on Poisson process. Next, it strictly derived the connective probability for the traveling vehicles on a road. In order to verify the correctness and validity of the two hypotheses and the proposed network connectivity model, it chose the traffic data in the period of 7:30 am to 8:30 am for the medium city Luxembourg in European as the experimental scenario. After that, it implemented the simulation on the urban traffic simulation platform (simulation of urban mobility, SUMO) . In the simulation experiment, it simulated the probability distribution of vehicle speed, the arrival rate of vehicles, and the average number of vehicles in the road and the probability of network connectivity. The experimental results show that the calculated values of the theoretical model are consistent with the simulation results. Hence, this paper proves that the two hypotheses and the proposed networks connectivity model are reasonable and correct.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-12 Cooperative journals: 《计算机应用研究》
Abstract： To solve the problem that recent research about multi label propagation community division algorithm adopted the random sequence strategy to result in unstable result of community division and poor community quality, this paper proposed a Multi-label Propagation Algorithm Based on the Node Comprehensive Similarity (MLPA-NCS) for community division. This paper chose the descending order of node potential impact as the node selection order in order to solve the problem of the instability of the propagation. Node synthesis similarity could be calculated based on the theme of node similarity and link correlation, and it’s descending order was used as the order of neighboring nodes traversal when updating the node label to improve the quality of the communities found. This paper used real data sets and artificial network data to compare the results of several algorithms. The results show that the algorithm is effective and feasible and able to make the result of community division more stable while the quality of community more effectively.
Hits
Hits
Downloads
Comment