- EI核心期刊
The article attempted to solve the challenge on how to define and qualify constraints to get solution when making migration strategy. First, the constraints were classified, quantifiable constraints were calculated. Then the analytic hierarchy process was used to change quantifiable constraints into numeric importance, and the best solution was made. Last, the hybrid migration strategy for each component of a service operation support system was applied.
Based on modified Carreau constitutive equation, the Reynolds equation was deduced, which considers shear-thinning non-Newtonian property. The flow curve of poly alpha olefin (PAO)650 was measured with high-pressure rheometer. The rheological parameters were obtained by fitting the related test data. PAO650 elastohydrodynamic lubrication (EHL) film forming properties were studied. Experiment indicates that PAO650 film forming properties based on shear-thinning EHL theory very well agree with the experimental results. The shear-thinning non-Newtonian rheological effect makes PAO650 film-thinning severely. For PAO650, the entrainment rate index of classic EHL film thickness formula needs to be adjusted. The slide-roll ratio has little impact on PAO650 film-forming properties.
The secrecy rate of the source-destination link can be effectively improved with help of friendly jammers, but the jamming service is not free. Obviously, there exists a tradeoff between the friendly jammer and the source-destination pair. An optimal power allocation algorithm based on Stackelberg game was proposed. Based on the proposed two-level Stackelberg game model, the existence of unique globally optimal energy efficiency was proved, and the closed-form result of the optimal energy efficiency was derived. Simulations show that the proposed collaboration strategy is better than the average power allocation strategy in terms of energy efficiency. And multi-nodes collaboration could achieve better performance.
Aiming at the parameter estimation of the fast frequency hopping signals for underdetermined situation, an improved blind parameter estimation algorithm was proposed, considering the hopping instants detection method based on autoregressive moving average (ARMA) model. The parameter estimation of the fast frequency hopping signals was realized by modifying the time-hopping sequence according to the hop duration and Fourier transform. Experiments show the estimation accuracy of the algorithm can reach more than 90%, about five times increasing compared with the existing algorithm for underdetermined situation when the signal to noise ratio is greater than 10 dB.
In order to balance the load of Hadoop distributed file system with lower load migrating cost and data transmission cost, a certainty multi-stage and multi-object (CMM) decision model was proposed. The model is a directed acyclic graph built on decision nodes, which adopts the remaining load capacities of CPU, memory and disk as decision preconditions, and also adopts load balancing effect, load migrating cost and data transmission cost as decision targets. By CMM model, the best balancing plan is determined by selecting results of multiple decision stages and computing the plan usage. Simulations show that the CMM based strategy can achieve better load balancing effect, load migrating cost and data transmission cost.
In order to guarantee high reliable service in optical burst switching(OBS)networks and reduce the end-to-end delay of bursts, an online burst encoding mechanism based on irregular repeat accumulate(IRA) codes was proposed. The parity-check matrix of IRA codes was adjusted into a special structure, and then an on line IRA (OL-IRA) code which can encode and decode burst online was constructed. Information data bursts are encoded by OL-IRA codes to recover lost bursts with redundant bursts. Simulations show that OL-IRA codes will guarantee the performance of burst loss recovery, and can also shorten the delay of burst loss recovery.
Aiming at the problems of security situation awareness about networks, a network security situation awareness method based on neural network is proposed. Firstly, an index system of network security situation evaluation was built and the network security situation was defined with four independent properties. Then, a back propagation (BP) neural network based on the method to estimate the security situation assessment of network was designed. To solve the problem of uncertainty and fuzziness between the situation factor and the evaluation result, a network security situation forecast method that can find non-linear mapping relationship among network situational values was given based on radial basis function (RBF) neural network, parameters of the network were optimized through adaptive genetic algorithm and the network security situation awareness was obtained. An experiment with a real network environment was performed, the experimental results prove that our method is feasible and effective to the network security situation awareness.
In order to eliminate the ambiguity of named entities in the documents, a named entity disambiguation algorithm combining entity linking and entity clustering is proposed, and the proposed algorithm combines two methods to compensate for the limitations of only using one of the methods. The proposed algorithm expands the mentions in the background document firstly, and generates candidates in the English Wikipedia knowledge base for expansions secondly, then extracts a variety of features to rank candidates, lastly uses clustering to disambiguate the mentions which has none candidates in the knowledge base. The experimental results show that, in the proposed algorithm, the F measure in KBP2011 data set is 0.746 and the F measure in KBP2012 data set is 0.670.
Aiming at solving the problem that information about the distribution of network traffic load cannot be correctly captured by interferer neighbors count (INX) routing metric, an interference and regional load aware (IRLA) routing metric for wireless mesh networks was proposed. Average competition degree was used to describe the competition degree for the same channel among interferer links and the dispersion degree of traffic load on interferer links. The information was utilized to guide the routing selection for avoiding heavy traffic load regions and achieving network load balance effectively. Analysis and NS-2 simulation show that the proposed routing metric can not only improve network throughput dramatically, but also reduce end to end delay and packet loss ratio, and it can even enhance the network performance in heavy load environment.
A new 8-antenna double codebook design scheme was proposed to improve TD-LTE-Advanced system's spectrum efficiency. Therein, the codebooks for differently layer number were constructed according to full rank matrix provided in Rel-8 protocol or discrete Fourier transform matrix combining with cyclic delay diversity. Analysis dedicates the new codebooks have larger intervals in codewords comparing with the 8-antenna double codebook in Rel-10 protocol. Simulation shows that the proposed method can obtain higher gain at least 2 dB for 8-transmitting and 2-receiving antennas and 2 layers; while the new codewords can increase spectral efficiency more than 15% compared with existing codebooks when signal to noise ratio is 5 dB.
The anti-collision technology is a key technology in the radio frequency identification (RFID) system. In order to avoid the low efficiency problem of reading massive data labels via the binary-tree search algorithm, by analyzing the character of mass data and the regularity of tags encoding, a flat-N-tree search anti-collision algorithm is proposed according to the regularity of massive data tags encoding and the principle of binary-tree search algorithm in the management systems, which is suitable for reading massive data RFID labels. The result shows that the reading efficiency of the multi-tree search algorithm is better than that of the binary-tree search algorithm when the total quantity of tags is greater than a certain value.
Inspired by principle of multiple thinking system of coordination of brain, a model named as pragmatic coordinative model was proposed. Pragmatic information is used to govern the coordinative procedure of analogue image thinking system and analogue logic thinking system. The pragmatic coordinative model has large capacity of pattern recognition and fault tolerance. Consequently, a background model based on the pragmatic coordinative model was constructed. The background model can represent the scene effectively and tolerate the typical disturbance.
A new forgery detection scheme is proposed in case of modification of images formed naturally. Illuminating an object leads to make diffuse refection on surface of the object, its shadow appears in the projection plane. Once images are tampered, it is difficult to maintain the consistency of these natural features. Therefore, the uniformity between the direction of light estimated with an object in a photographic and the area of the light direction on the basis of cast shadow, will not effectively work for a detected image of multiple objects, but determine if an image with only one is tampered or not.
Aiming at the class labeling starvation and class imbalance problems in Internet traffic classification, a bootstrapping based traffic classification method was presented. An initial classifier was trained on a small number of labeled samples, and then it is updated iteratively by predicting the class labels of unlabeled samples and extending the training set. A new algorithm was devised to compute the confidence used for selecting new labeled samples into the extension set. It correctly adopts classifying unlabeled samples with a posterior probability distribution as probabilistic event and to decrease the noise in the extension set. Moreover, the heuristic rule was built with aid of probably approximately correct theory, its biases is toward selecting minority class samples into the extension set so as to reduce class imbalance degree. Experiments show that the bootstrapping based classifier gets improved of 9.46% on overall classification accuracy compared with initial classifier, and the recalls of minority classes get increased about 2.22% averagely compared with the existing method.
A green grooming algorithm, called as time-aware green grooming (TAGG) algorithm, is proposed for scheduled traffic requests. By comparing the transmission energy consumption for different traffic grooming strategies, the most energy-efficient traffic grooming strategy is selected in TAGG algorithm for traffic requests. Moreover, in order to further reduce the transmission energy consumption, a scheme is put forward to trigger the operation of merging light-paths. According to the moment of tearing down light-paths, the scheme can determine the moment of merging light-paths for traffic requests, which can reduce the number of required router ports and transceivers, and improve the wavelength bandwidth utilization. Simulation shows that the proposed TAGG algorithm can reduce transmission energy consumption effectively.
As the storage system grows, how to improve the system reliability has become a key issue to the storage system. The redundant array of independent disk (RAID) codes of tolerating double failures can not meet the requirement of reliability in storage system. On the basis of the row diagonal parity (RDP) code for double toleration failures, a new class of extended RDP-RAID code for triple storage failures was presented. The three nodes failure recovery capability for a given data redundancy were optimal. The simple and intuitive algorithms of encoding and decoding were proposed by using binary matrix transformation. Analysis shows that the comprehensive properties of the proposed code are better than the STAT code and the EEOD code, such as update complexity, encoding and decoding complexity, storage efficiency. And it shows high reliability for storage systems.
In order to make the link sign prediction more accurate in signed networks, it is necessary to analyse each underlying principle of generating signed networks. Structure balance theory and status theory are extended to gain more information for link sign prediction. A new measurement named PageTrust in web network is introduced to describe the importance of node of signed networks. On the basis of integrating different kind principles of generating signed networks, a group of refined features are extracted. Based on those creative features, two link sign predictors using supervised machine learning algorithms are established. Experimental results on two real signed networks demonstrate that learned model is more accurate and generalized than other state-of-the-art methods.
In cloud computing, the data and computation migration gives rise to trust problems between the user and the cloud. Including the hardware platform, the multiple dimensions method was studied, as well as the user identity and behavior. The electronic identity (eID)-based virtual machine trusted attestation system in infrastructure-as-a-service (IaaS) cloud was designed. The hardware platform was used for trust third party architecture. The citizen's network eID was used as users unique authoritative identity. The credit records were also applied to evaluate the user's behaviors. Four steps were adopted to solve the trust problem between two sides, including trusted attestation of the user identification and trusted attestation of the virtual machine. Experiment analysis shows that this system can defend common attacks, it is more safety, and the time complexity is within acceptable limitations.
When the fanout number of multicast cells is relatively small, the existing cell assignment scheme has poor performances under balancing traffic load. The scheduling algorithm can almost not benefit from the increase of multicast queues. A weighted modulo (WM) algorithm for multicast cell assignment was presented, which can ensure more scheduling opportunities and work conservation. The proposed scheme performs modulo operation on the weighted sums of the fanout bits to get the mapped queue number, thus the multicast cells with small fanout numbers can be balanced very well. Simulations show that multicast scheduling algorithm adopting the proposed scheme has higher throughput than that adopting the Modulo algorithm.
In view of shortcomings of distributed coordination function (DCF) in IEEE 802.11 series standard, a simple but effective improved DCF mechanism was proposed. Main improvements are focuing on the size of backoff window and it's probability was chosen in 0 backoff stage: doubling the backoff window size of 0 backoff stage based on original mechanism and using probability p to divide the window size of 0 backoff stage into two parts. For purpose of reflecting the performance accurately, a comprehensive analytic model was constructed. We compare it with the classical DCF scheme and numerical simulation. It is shown that the new DCF mechanism has better performs than that of classical DCF in the case of dense terminal and heavy offered load.
Considering the characteristics of predictability for cyber physical systems (CPS) and the requirement of secondary storage for node systems, a research was carried for the existing flash file and sensor node file systems. A new node file system for CPS was designed, adopting virtual sector mapping technology and cascading page log strategy to ensure predictability, and also optimizing the file structures and data operations to improve efficiency. Experiments on the flash simulator show that the system has stable performances of sequence input/output (I/O) and random I/O. And the system image keeps small that means it is suitable to deploy on resource-limited nodes.
The conventional least mean square (LMS) based adaptive beamforming converges very slowly under low signal-to-noise ratio (SNR), thus an adaptive beamforming algorithm in wavelet domain with a fast convergence rate was put forward. The white Gaussian noise could be erased by means of wavelet transform soft-threshold method. Besides, Newton's method was further applied in LMS algorithm for beamforming in wavelet domain. Simulation shows that the proposed algorithm indeed improve the convergence accuracy and the convergence rate compared with either the conventional LMS algorithm or the existed LMS algorithm in wavelet domain. It also substantially improves the accuracy of beamforming compared with the conventional LMS algorithm.
For synchronization performance under imprecise antenna pointing condition, a satellite mobile communication system is established based on the geostationary orbit mobile radio interface (GMR-1) standard on the OPNET platform, the entire satellite synchronization performance is studied when the downlink antenna points inaccuracy. Its core idea is to complete the inaccuracy antenna pointing impact simulation through the interactive among mobile earth stations and gateway stations, to establish the mathematical model, to analysis and simulate the time synchronization and frequency synchronization performance to achieve the quantitative impact under imprecise antenna pointing condition. Analysis and simulation show that, compared with the satellite that has precise pointing antenna, the imprecise condition has little impact on the time synchronization, however, the frequency synchronization performance is relative to the location of the user, it may reduce the bit error rate performance by 10-1 to 10-2.
A virtual network reconfiguration(VNR) algorithms based on the minimum cost is proposed. In the method,achieve the reconfiguration mechanism, the parameter—the degree of improvement during the reconfiguration—is defined to judge the effect of network reconfiguration mechanisms, then the target physical nodes are selected on the constraint of the minimum cost, after then, the virtual node will be migrated to the target physical node, and then virtual links will be mapped to the physical path using the shortest path algorithm to calculate. This method will avoid the phenomenon called "seesaw" that the target physical node becomes the new bottleneck node. The reason of the phenomenon is that virtual node is migrated from one physical node leading to bottleneck to others during the VNR. The results of simulation verifies the feasibility of this method.