- EI核心期刊
It is introduced the real time data processing system (RTDP) network and its application fields first, based on these work the author proposed a new real time processing system network model. Then the author analyzed the hot research point about the frame work of RTDP network. Finally, it is described the future work of RTDP network.
A random virtual network embedding algorithm based on maximum independent link set was proposed. The algorithm redefines the concept of matching in the graph theory for the weighted graph and names it as independent link set.In order to improve the success rate of the virtual network embedding and reduce consumption of link embedding,in the stage of embedding virtual links in the independent link set, physical links are filtered by the availability of resources, and then the virtual link is randomly embedded to a single physical link.Randomly embedding can guarantee the load balancing for physical network. Simulation results show that the algorithm can effectively reduce the link consumption and improve virtual network acceptance rate.
In order to solve the issue that the individual driving force lacks in traditional opinion model, a dynamics opinion model based on expected attraction and trust neighbor mass(DOET) model-dynamics opinion model based on expected attraction and trust neighbor mass is proposed. It combines exception-traction and herding effect which can establish state transfer and three kinds of opinion decision. The opinion is formed through DOET model with considering two factors of internal expectation attraction and trust neighbor mass. Experiments show that this model can simulate unification, polarization and mild concussion of opinion under influence of exception-traction and herding effect; because of network structure and distribution of initial opinion, the "Compliance" node is not disappeared but kept in existence. Also, DOET model is correspond to the characteristics of opinion formation and individual interaction with comparing existing opinion model. Consequently, this model can express internal disciplines also can be referred to the network analysis in opinion formation.
Research on economic theory to solve distributed resource allocation problem, especially its characteristics in distributed computing is ignored. The difference between two environments is analyzed. It is pointed out that the key elements of ignored features in distributed computing is resource, task, individual and price. The framework in neoclassical economics to make economic analysis for distributed resource allocation is given by comparing the resource allocation process in distributed market environment. The modeling method of how to use economic theory to solve distributed resource allocation is described.
The main role of wireless sensor networks is to collect environmental data. As for the sensor nodes are vulnerable and work in unpredictable environments, the sensors are possible out off work and return to unexpected response. Therefore, fault detection is important in wireless sensor networks. The authors propose a fault detection algorithm based on support vector regression, which predicts the measurements of sensor nodes by using historical data. Credit levels of sensor nodes will be determined by a contrast between predictions and actual measured values. Then the dependable data set which is constructed by high credit level measurements will be used to detect sensor faults. Simulations demonstrate that the algorithm works very well in conserving energy and raising failure detection rate.
In order to reduce the influence of the wearing locations of mobile phones for user behavior recognition, a coordinate transformation method was proposed. The acceleration data acquired from the mobile phones' embedded accelerometers was transformed from mobile phone coordinate to the azimuth coordinations by means of the azimuth signals from the embedded direction sensor. Ant colony algorithm was used to choose the sensitive feature sets from the different coordinates' acceleration signals. BP neural network was employed to recognize the mobile users' behaviors. Experiments show that the proposed coordinate transformation method effectively reduces the influence of gravitational acceleration and wearing locations, and can observably enhance the recognition accuracy of the mobile users' behaviors as well.
Game theory was applied on network defense very well. And static model was used widely in most of the previous studies. However, there shows that such models have not take attack-defense cost into consideration and also cannot follow the evolving of the intention and strategies of attacks. A stochastic game model is proposed based on host vulnerability information. Combining host important degree and success rate of defense measures, attack-defense cost in single security attribute is analyzed, then a cost quantitative method with attack-defense intention is provided. An algorithm for attack-defense equilibrium strategy selection based on those models is proposed. Another example is given to illustrate the efficacy of this model and algorithm.
An assessment method of users' quality of experience (QoE) for mobile Internet based on transmission control protocol (TCP) packet layer analysis is proposed so that the metrics could be measured on the edge node of the mobile internet. This method gathers information from users' behaviors and perceptions, which could be applied to a single subscriber granularity at its best. While doing assessment in users' behaviors, the page refresh incidents should be paid more attention, which is analyzed by the TCP flags of the interaction sequence. On the other hand, users' perceptions indicators could be achieved using three different methods according to various requirements spanning from a coarsest grain perspective to a finest grain perspective, it would be later matched into a QoE mean opinion score (MOS) using the exponential independency model (IQX) of QoE and quality of service.
The delay performance of mobile Ad hoc networks with two-hop relay algorithm is analyzed. Computing the mean transition steps of the geometric random process and joint probability distribution of random variables, the closed-form expression of the delivery delay upper bound and the optimal transmission redundancy strategy are obtained. Further, based on mean residual service time, the end-to-end delay performance is derived. Finally, the numerical results show the relationship of delay performance with the network size and the traffic load.
A method of key node ranking for road network based on tripartite graph was proposed. The statistical information of routes and origin-destination(OD) of trips were extracted from taxi trajectories, and then a tripartite graph was built to model interrelationship among the nodes consist of trips, paths and intersections. The ratings of these nodes were synchronously calculated in an iterative process. This method not only incorporates topological structure and traffic characteristics, but also takes correlation between intersection and OD distribution into account, so network-wide key nodes can be identified accurately. Experiment verifies the validity of the method.
Because the extremely high availability and the characteristics of risk aversion are demanded in electric power communication network, there is great difference between the electric power communication network and conventional communication networks. The minimum loss of credible parameter (MLCP) algorithm has been proposed for the problem that the missing faults can lead to the possibility of power accident. Briefly reviewing and summarizing the symptom of coverage and contribution in classic algorithm, the impact of the fault loss is put more attention to, which is of advantage in power communication network and add the credible parameter to select the hypothesis fault firstly. Simulations show that MLCP algorithm can make the reasonable explanation for the observed symptoms with smaller false positive rate (FPR) and averaging running time of algorithm than other algorithms, without increasing the complexity of the algorithm. The MLCP algorithm is more appropriate to the fault location in the electric power communication network.
Because traditional duplicate image detection technologies can not ensure the scalability and the precision of image retrieval, this article proposes a duplicate image detection approach based on two-dimensional cloud model filter. On basis of bag-of-word model, the approach first maps the matching descriptors which are refined by Hamming embedding to points in the two-dimensional space, and then uses cloud model to compute the uncertainty of two-dimensional points' distribution for excluding the candidate images with larger fluctuation. Finally, the images are ranked according to voting score. Experiments show that the new approach not only maintains the merit of weak geometric consistency constraint algorithm which is suitable for large-scale image retrieval, but also significantly improves the accuracy of duplicate image detection.
To evaluate the transmission capacity of vehicular Ad-hoc networks (VANETs) wireless link, a link transmission performance evaluation model used for vehicle mobility and data transmission character is proposed. VANETs link transmission capacity metric is built based on link residual lifetime distribution and data transmission service time distribution obtained by Monte Carlo simulation and data fitting. The evaluation model is applied in VANETs simulation to assess the transmission performance of paths selected byad hoc on demand distance vector (AODV) protocol or prediction based on routing (PBR) protocol. Simulation is carried out on TrueTime tool box of Matlab/Simlink and is shown that the link transmission performance evaluation model is able to assess link transmission capacity effectively for both route protocols and different node speed scenes in data transfer service.
For the Aloha based anti-collision algorithm in radio frequency identification networks, tag collisions could greatly reduce the throughput of the system. If the number of tags was got, the throughput could be greatly improved. Based on maximum likelihood estimation, the proposed hybrid tag number estimation scheme combines the binary search based anti-collision algorithm and aloha based anti-collision algorithm to estimate the tag number. Simulation shows that the proposed scheme has higher estimation accuracy than the existing algorithm.
A method of dynamic intrusion detection using vaccination (DIDV) is designed. With rough set (RS), a scheme is given to generate antibodies, and a strategy based on the significance of the attributes is proposed to get vaccinations. The RS is used to promise excellent antibodies and to increase the detection rate. Vaccination strategy is applied to gain more compatible vaccinations and to get higher convergence rate. Experiments show that the proposed method is of feasibility and effectiveness.
In order to enhance the efficiency of resource utilization in cloud computing environment, the problem of optimizing the server selection process in a global view is focus on. A global optimized server selection scheme is proposed. A new parameter is introduced in terms of inter-domain transit traffic penalty coefficient to balance the proportion of data transmission between different internet service providers(ISPs) and network distance as well as between terminal and server. A scalable network distance prediction algorithm is given to reduce the cost for measuring network distance. The proposed scheme satisfies the needs of the users to improve the service experience, implements the load balance between servers and reduces the inter-domain transmission. On the other hand, the measurement cost is reduced in the process of server selections. Simulations show that the proposal improves the efficiency of the server selection, reduces the measurement cost and is with good scalability.
The topology control problem is considered, creating an energy-efficient topology of wireless ad hoc networks in presence of selfish nodes. A non-cooperative game framework is established to describe the interaction of nodes in topology control process, where each node tries to transmit minimum power to preserve a connected network. Some characteristics of Nash equilibrium topologies are analyzed. A distributed topology control protocol restricted information exchange among neighboring nodes is proposed. Simulations show that the game based protocols observably eliminate the redundancy of the initial network topology.
A security evaluation based on network entropy and stochastic game was presented. An attack defense stochastic game model was proposed to describe the conflict of network security which is dynamic and multi-state. The concept of network entropy was introduced to describe network security performance. The optimal defense strategy and network state probability was obtained by solving the Nash equilibrium of attack defense stochastic game model. On this basis, the security of network was assessed combining with network states entropy difference. Network security evaluation algorithm was given at last. An example is representatively provided to show that the method can effectively assess the security of network and predict intrusion behavior.
A named round-trip programming algorithm for estimating end-to-end delay was proposed based on the known linear programming algorithm in which the forward and backward delays are simultaneously measured. It regards the sum of forward and backward delays of different packets as one packet's "round-trip delay" in order to eliminate the influence of offset on end-to-end measurements. And if the packets in two directions, which experience least end-to-end delays, are sent at different time, this "round-trip delay" would be affected by skew. The algorithm adjusts one packet's sending time to another to avoid the effect of skew. Furthermore, the mentioned algorithm is compared with linear programming and Paxson's algorithm. Analysis and simulation prove that the algorithm is of better performance without increasing complexity.
Consisting of nDroidC (client) and nDroidS(server), a behavior-based Android malware analysis system: nDroidAS is proposed. Application installation events on the Android device are monitored by nDroidC, which generates analysis requests while an application is to be installed. The target application is installed in nDroidS, by which dynamic feature vectors of the application are collected and analyzed to detect the malicious ones. Meanwhile, to pre-analyze applications, an Android package(APK) fetcher is designed in nDroidS to fetch APK samples from app markets. Some key technologies of the system such as feature vectors selection and interaction simulation are also discussed. A simplified prototype of nDroidAS is built, which is able to analyze Android malwares dynamically and fetch APK samples in the wild. Experiments show that the proposed system architecture is feasible.
A new algorithm is used to simulate the coverage of base station,which shows both the coverage radius of base station and the location of the base station and fit in with the requirement of network optimization industry.New algorithm produce simulation results as fast as Voronoi and as same as circles intersection reflecting coverage radius of base station.The algorithm avoids dealing with the complex cases in which several circles intersect with each other by transform circles intersection as circle intersecting voronoi.Compared with the previous algorithms, new algorithm can greatly reduce the difficulty and computational complexity.
The prediction of job running time and computing resource is very important in production environment. However, the performance of cloud computing is not easy to be predicted due to complicated computing resources in environment of cloud platform. A model is proposed based on resource consumption patterns for predicting the running time and central processing unit (CPU) resources consuming of MapReduce job in cloud computing environment. This model comes from polynomial regression modeling to predict the performance of MapReduce job. A variety of criteria is used to evaluate the model. Experiment shows that this model could predict the running time and CPU resources consuming of MapReduce job with high accuracy.
A network resource personalized recommendation method based on K-means clustering algorithm is presented for dynamic multidimensional social network. Firstly, the user is modeled according to the user rating data, and a multidimensional network is constructed by collecting all the users' rating data, and then a dynamic multidimensional network could be formed with the help of local world evolving network model. Secondly, the network users are clustered by using the improved K-means algorithm. Finally, the objective user's rating could be forecasted and obtained by referring the nearest neighbors, and the personalized recommendations could be made. So far, a network resource personalized recommendation method suitable for dynamic multidimensional social network is formed. The experimental results show that the new recommendation method could reduce the error between the prediction value and the true value by comparing with the collaborative filtering recommendation system, and hereby, the new recommendation method could achieve the improved personalized recommendations.
In order to solve the poor network stability problem and the short survival issue which is caused by the separation between trust layered structure and physical clustered structure, a mobile Ad hoc networks (MANET) clustering model which based on trust measurement was proposed. The distance measurement of intuitionistic fuzzy sets were researched, and a trust measurement which using intuitionistic fuzzy set theory was proposed. A clustering algorithm based on trust measurement was proposed via analyzing the stability and energy dissipation of the clustered structure in extending HELLO message form. The simulation results show that the proposed model has an advantage in head nodes changing rate, the clustered structure member energy dissipation, as well as, MANET survival time.