A new packet forwarding scheme which is suitable for the length variable packet switch is proposed. In this new scheme, a length variable packet is chipped and forwarded as several fragments, i.e., the basic unit of the forwarding is the fragment. This mechanism can ensure that the short packet in the switch fabric is delivered as soon as possible, thus reducing the average packet delay and delay-jitter and improving the throughput of the network. Theoretical analysis and simulation results show that with the fragment-based scheduling mechanism, compared with the packet-based scheduling, the performance of throughput can be improved by 10% to 70%, and that the delay can be reduced by 10% to 30%. Moreover, the higher the proportion of short packets in the input traffic, the more remarkable improvement is obtained.
In order to improve the overall performance of medical image non-rigid registration, this paper presents a new medical image non-rigid registration method based on the LBM (Lattice Boltzmann Method). Firstly, this paper transforms medical image registration into a viscous fluid physical problem to take advantage of the fact that fluid particle movement can simulate any free complicated deformation. In the registration process of this paper, floating image, the gray level difference between the template image and the floating image, and image pixels are considered as a viscous fluid, external force and fluid particle, respectively. The external force from the gray level difference between the template image and the study image drives registration. Through simulation by use of the LBM, the flow of template image pixels under the external force which make the template image generate displacement field and deformation is obtained. When the gray scale values between the study image and the template image are near agreement the image registration has finished, at which time the external force disappears and the fluid flow stops. The LBM equation of the compressible viscous fluid with the external force is set up. In addition, the boundary conditions, initial conditions and speed module value are defined in this paper. Experimental results show that the non-rigid medical image registration performance has been improved comprehensively and especially the efficiency of image registration has a qualitative leap. This method is very suitable for large deformation registration occasions.
Under the assumption of the two-user Gaussian interference channels with a single Gaussian codebook, no power control and no significant coordination between the transmitters, a decoding strategy is proposed to achieve the best rate region or sum rate according to the channel gains and power parameters. The proposed decoding strategy exploits two fundamental decoding methods and is described in two equivalent forms of channel parameters, which shows that the best decoding strategy is not determined by the intensity of the channel gains but depends on both channel gains and power parameters.
For the problem of monopulse angle estimations of multiple targets based on the large phased array, a method for estimating bias angles using sum-difference multi-beamforming at the sub-array level is proposed. Firstly, system complexity is reduced by combining the outputs of sensors; secondly, sum-difference multiple beamformings are achieved by phased-alone sub-array steering vectors with the anti-symmetric mode; finally, the 2-dimensional bias angles of targets in arbitrary steering beams are estimated based on the two slopes along azimuth and elevation axes pointing to the normal. Compared with the method of four steering sum-difference beamforming, the proposed method has the advantages of higher accuracy and lower complexity. The measured and simulated data show the effectiveness and superiority of the proposed method.
A systematic study of the square split ring resonator (SRR) which includes the microstrip SRR and SRR Defected Ground Structure (SRR DGS), and an analysis of the resonant and coupling properties of the SRR have been made. Calculation formulas for the coupling coefficient have been obtained by analysing the equivalent circuits. The coupling intensities between SRRs at different positions are analyzed, which has important meaning for filter design. Finally, two bandpass filters are designed using different coupling methods.
An incremental clustering algorithm is proposed to identify community structures in dynamic networks. Based on the feature that in dynamic networks there is little change in adjacent network snapshots, the community structures detected in last snapshot are used as the initial clustering results in current snapshot. Then the edge bridgeness is adopted to judge the snapshot change's influence on clustering results. Finally, the community structures fitting current snapshot are obtained by locally modifying the initial clustering results. The accuracy and efficiency of our algorithm are validated by comparing with the MCL algorithm. Experimental results demonstrate that our approach performs accurately and effectively in identifying community structures in dynamic networks because clustering the current snapshot can be avoided by incrementally analyzing the dynamic networks.
A new network model is proposed for efficient analysis of crosstalk when tracing the cross split return path plane. The influence of the non-ideal return path plane is described as coupling between slot line mode and microstrip line mode, and then circuit structure is decomposed into the slot transmission line and microstrip transmission line which are connected by an ideal transformer. According to the characteristic of the cascaded network, the transmission matrix for the 4-port network is selected to calculate crosstalk. Comparing with full wave simulation, the analysis time is reduced from 60min to 30s with considerable accuracy. The shorting lines which cross the slot in the return path plane are used to bypass the slot line mode in order to improve transfer performance and suppress crosstalk. Simulation results and experiment test show that the near end crosstalk and far end crosstalk are decreased by about 25dB and 20dB, respectively. Compared with the traditional decouple capacitor method, the shorting line proposed in this paper has benefits of wide bandwidth, easy layout and low cost.
To solve the problems in information security risk assessment, such as inaccurate security classification and long assessment time, a risk assessment model of information security (RAMIS) is proposed based on Support Vector Domain Description (SVDD), and is called SVRAMIS for short. Firstly, SVDD is applied to obtain the minimal enclosing ball (MEB) of each class, and disconnect regions are obtained by the description boundary. Secondly, based on the information provided by the hypersphere centers and the hypersphere radius, the positions of the test samples are confirmed, so that corresponding discrimination rules can be adopted. Finally, numerical experiments on information security data demonstrate that, for various kernel functions, the proposed model can lead to high training and testing accuracies and short training time.
Previously proposed dispatching schemes for MSM Clos-network switches lack an efficient support for multicast forwarding. In this paper, a novel MSM Clos switching fabric and a simple scheduling scheme, called Multi-/Unicast Static Round Robin Dispatching (MUSRRD), are proposed. The new architecture, called the Multicast supporting Clos-network (MClos), allocates a small number of Virtual Multicast Queues (VMQ) at each input module for multicast traffic. Each slave arbiter is not only required to select which cells are to be transimitted but also needs to decide from which set of queues (VOQs or VMQs). MUSRRD uses a simple static round-robin arbitration scheme to schedule both traffic types, which is fair and easy to implement. Simulation results show that MUSRRD can achieve better performance of throughput and delay than the alternative solution under different traffic patterns.
Based on the MMSE frequency domain equalizer, channel estimator, and convolutional decoding, an iterative receiver in the time-frequency domain(TFD) for SC underwater acoustic point to point communications is developed. This iterative procedure can reduce the calculation complexity of equalization using frequency domain processing and obtain better performance using less receiving elements. Finally, we use the sound speed profile(SSP) measured in the lake and finite-element ray tracing(Bellhop) to simulate the underwater channel to verify the performance of the proposed receive algorithms.
A combination optimization routing algorithm (CORA) based on the ant colony optimization in dual-channel wireless sensor network is presented to put down the blocking probability of high load network. This algorithm deals with the date collision and multicast suppression in the channel competitive process well by the dual-channel communication model. At the same time, this algorithm uses the infection sphere to reduce the number of nodes which join in researching the optimization route from the source node to the target node, and thus can reduce energy consumption of the network. Finally, this paper proposes a combination optimal routing algorithm with a layered-graph model. The service blocked in the control plane can use the idle resource in the data plane for transmission in a synchronous manner, so the blocking probability of networks and the delay of communication can be cut down in this way. Simulation results show that this algorithm performs better in terms of the time consumption of communication and the total energy consumption. The blocking probability of networks can be cut down 13% compared with the EEABR and EEAWSN.
For the low efficiency and poor accuracy of the resonant cavity transmission line model based on the double frequencies approximation, this paper proposes a fast algorithm based on Pade approximation and gives the corresponding fast and accurate Spice-compatible resonant cavity transmission line model. Compared with the double frequencies approximation using the single LC resonant circuit, Pade approximation can adjust flexibly and improve the efficiency and accuracy of the model due to using multiple LC resonant circuits. Simulation results show that the efficiency of Pade approximation is higher than that of the double frequencies approximation, especially when the number of transmission lines is great, and that the error of Pade approximation is less than 0.007% and is better than the error 0.3% of double frequencies approximation.
A new method named the predicate-membership logic representation method is proposed in this paper, which is based on the facts that the predicate logic can represent certain knowledge and the membership function in the fuzzy logic can represent the uncertain part. The knowledge base of PCB technique rules and the knowledge reasoning model are achieved by normalizing two-valued and multi-valued logical rules with the predicate logic of the membership function. Meanwhile, they provide the examination basis and data support for the intelligent examination of PCB.
One of the important goals of the developing antenna array of solar imaging is how to decrease the cell number of the array, due to the contradiction between limited budget and massive antennas. Inspired by the ability of compressed sensing to recover exactly the original signal from highly sub-Nyquist-rate samples and the sparse characters of solar image, this paper proposes a solar image reconstruction method based on the compressive sensing theory. The effectiveness of this scheme is illustrated by experiments of both simulated solar image and real astronomical image. Results show superiority of the proposed method in the resolution of the adjacent point source, the shape maintenance of the extanded source and the dynamic range.
To improve the low resolution of images reconstructed by conventional free view computational reconstruction methods, a free view computational reconstruction method based on the light field model is proposed. High resolution free view integral imaging computational reconstruction is realized by using the light field projection transformation matrix to project the elemental image array to the reconstruction plane according to the view point and viewing direction, and thus the information in the elemental image array is fully utilized. Simulation results show that the proposed method can achieve free view integral imaging computational reconstruction, and the resolution of the images reconstructed by the proposed method is improved greatly compared with the conventional method.
The cylinder perspective projection model is proposed to correct the image distortion according to the imaging feature of the PAL. In the cylinder projection, model a line in the space is not projected to a line but an ellipse whose short axis is constant. Firstly, the PAL correction model which includes the distortion parameters is constructed. The points on the space line are projected to the points on the cylinder under this model. All spherical points are fitted to the best great circle under the condition that the sum of projection lengths of all points. The distortion parameters are obtained during this procedure. Then the panoramic image is corrected. Simulation and real image experiments demonstrate that the correction results of tangential and radial distortions appear satisfactory. The error is controlled at the sub-pixel level.
The wireless sensor network(WSN) is featured by no infrastructure, distribution, resource constraints and limited processing and memory. Accordingly, in designing the time synchronization protocols for wireless sensor networks, intensive computation and complex route selecting are undesirable. For accelerating the time synchronization and lowering energy consumption in the WSN, this paper presents a simple time synchronization scheme for wireless sensor networks. In this scheme, each note broadcasts its clock information and as a result its neighbors will receive the clock information. After averaging the received clock information, the neighbor notes take the averaged clock information as its next clock tick. This process is carried out repeatedly until all the net notes meet the same clock tick, which means the whole network achieves distributed synchronization. As each note in the network only receives its neighbor's information, so this scheme does no need specific routing and too complex processing and it has a fast convergence rate and low energy consumption. The proof for the convergence of the proposed synchronization algorithm is carried out using the random matrix theory. The analysis of the convergence rate and the energy consumption and synchronization error are also carried out. The results of theoretical analysis are verified by computer simulation.
There are two square-root terms in the range history of a return signal from a Bistatic Synthetic Aperture Radar (BiSAR). The transfer function for imaging in the 2-D frequency or range Doppler domain using the principle of stationary phase cannot be analytically derived. To address this problem, we approximate the stationary phase of the 2-D spectrum with an expansion of the Taylor series at the azimuth frequency, and call the approximation the derivatives of an implicit function. After algebraic manipulation, the 2-D spectrum is obtained for an azimuth invariant BiSAR. With the proposed method we dissolve one square-root term out of two for an azimuth invariant BiSAR, which is especially advantageous in implementation of an imaging algorithm. Then, a modified Range Doppler Algorithm (RDA) is developed to process the BiSAR data. The promising results of simulation and real data processing are obtained.
A new algorithm is presented to solve RS code blind recognition problems. By searching the code length and the field parallelly based on the parith check function of the primitive element, the recognition efficiency is improved. The reliability of roots' searching is promoted by omitting error contained code words which do not accord with the primitive element parith check. The generator polynomial is forward-backward searched using the continuity of roots to simplify calculations and accelerate the search speed. Simulation results indicate that the new algorithm's upper limit of BER when correct recognition rate is 90% has a significant increase.
A multi-class classifier for feature fusion is designed based on the traditional linear Relevance Vector Machine (RVM). The proposed classifier extends the binary RVM to multi-class RVM based on the multinomial Probit regression model, utilizes the feature selection property of the linear RVM to reduce the dimensionality of the fused feature vector, and makes the linear RVM framework have the ability to form the nonlinear classification boundary via the rational power extension. All of these properties can ensure the robust fusion recognition performance in the case of nonlinear multi-class classification. For the application of radar target recognition using the high-resolution range profile (HRRP), the experimental results on the measured data show that the proposed multi-class classifier with the three translation-invariant features extracted from HRRP data can achieve robust recognition performance.
The existing geomagnetic matching technique demands information from the inertial navigation system (INS). In order to reduce the cost of the navigation system, a novel matching method with both the magnetometer and milemeter is presented. To search the path in the huge space, the genetic algorithm is employed. Correlation measure is used as the fitness function and the random variation is superseded by oriented variation. Simulation results show that this approach can be used for effective matching in different forms of path with enough positioning accuracy.
Particle filtering (PF) is particularly useful in dealing with the blind channel identification and blind equalization for its fast convergence and its outstanding performance of resisting multiple-path fading channels. Considering the Markov chain property of convolutional codes, the signal model is modified and a particle filter algorithm for joint blind equalization and decoding of convolutional code is introduced which samples the information sequence directly instead of the coded sequence. An iterative method to approximate the noise power is proposed, which is applied to the joint algorithm to adjust the parameter of noise power adaptively. The proposed algorithm is simulated. The simulation result shows that the convergence of the joint algorithm is faster and the bit error rate (BER) is lower that of the separate algorithm. And the adaptive adjustment algorithm reduces the computational complexity.
This paper proposes a pseudo-range differential positioning method based on the existing mobile phone base stations to improve the point positioning accuracy in the Chinese Area Positioning System (CAPS). The interpolation data for pseudo-range error correction in the generalized extended interpolation model are derived from the pseudo-range error value from several base stations to obtain the measuring point of the corrected pseudo-range error, and those data will be broadcasted to the user for pseudo-range differential correction. Simulations indicate that this method can overcome the fall in accuracy caused by the increase in base line length in conventional pseudo-range differential correction. The combination of the CAPS system and mobile phone base stations can effectively utilize the existing resources in mobile phone base stations. Moreover, it can also significantly improve the positioning accuracy of the CAPS.
Functional Decision Diagrams (FDDs) are a graphical expression for Reed-Muller (RM) expansion, and its variable order and polarity jointly influence the delay and area of the corresponding circuit. By studying FDDs and fixed polarity RM (FPRM) expansion, a delay and area optimization algorithm is presented. Firstly, a delay evaluation model is set up by FDDs. Then by combining the tabular technique and the search strategy of variable order, the optimal polarity and variable order are searched for both medium-scale and large-scale circuits according to the delay and area. Finally, the algorithm is tested with MCNC Benchmarks. The results show that the proposed algorithm is very effective for delay and area optimization.
Due to the intersymbol interference (ISI), the Maximum Likelihood (ML) algorithm using the cyclic prefix (CP) in orthogonal frequency division multiplexing (OFDM) systems performs poorly in frequency selective fading channels. Firstly, a novel timing metric is presented, which avoids the disadvantage of ensemble correlation. Using the sample data that is not affected by the ISI, the novel algorithm performs correlation operations between two data blocks which are a distance of an OFDM symbol length apart from each other. Secondly, we propose a novel detection function. By searching its maximum in a search range decided by the maximum of the timing metric and the length of CP, the timing offset is gained and the time synchronization can be obtained. Simulation results show that the novel algorithm has more accurate time offset estimation, smaller Mean Square Error (MSE) and lower complexity in frequency selective fading channels.
According to state estimation for the nonlinear system with intractable inequality constraints, a novel constrained unscented kalman filter (CUKF) is proposed. The objective function is derived by using the maximum probability method, inequality constraints are treated skillfully as a penalty function, and the optimum constrained solution can be solved iteratively using the adaptively step length method. Through theoretical analyses, the constrained solution is the rigorous local minimizer of the objective function. A target tracking example based on digital navigation is presented to illustrate the efficacy of the CUKF. Simulation results show that the CUKF has a better filtering accuracy.
Based on linear block codes and convolutional codes, the paper constructs a (2k, k, 2) convolutional code, with the state transition described by defining a 3d matrix. In the decoding part, some matrix processing modules are brought in to design a viterbi matrix decoder with parallel processing capability, and the single structure of decoder is good for its analysis and design. Simulation results demonstrate that this convolutional codes really possess a highly-efficient decoding speed and good error-correcting capability.
The development of the belt-type sensor networks derives from the development of the wireless sensor network. In the most general sense, the belt-type sensor network is a sensor network in a special form. The application requirements and prospects of the belt-type sensor network are extremely wide in many important fields. For the effective coverage problem of the node in the belt-type network region, the quantitative mathematical modeling analysis of node coverage is implemented, and a self-deployment algorithm DSDA-VC for the node distribution of the Voronoi cell unit is designed. The method can effectively improve the deployment coverage for the node of the belt-type network. By simulation check, the coverage rate of the belt-type network with a relatively high coverage density can be increased by 10% or more. According to the results, the design and actual validation test of the belt-type network in a mountainous area are carried out.
The Intersecting Cortical Model (ICM) possesses the Autowave nature stemming from the connection function during the firing process, but poses a problem called interference, which could blur the edge and detail in image processing tasks. Combined with the advanced development of vision biophysics, the paper makes an in-depth study of the solution based on the construction of the Centripetal Autowave and points out problems of the CA put forward by Kinser. Then, two new ways to conduct CA developing from the curve evolution idea are proposed, respectively, based on the linear heat flow and morphological median set, which truly solve the interference caused during ICM firing iteration.
By investigating the two-hop cooperative cognitive network in an overlay sharing(OS) mode, a cooperative detection strategy for the multiple cognitive user system is considered and outage probability formulation on the Amplify-and-Forward(AF) protocol is deduced. By optimizing the parameters such as sensing time and threshold, an optimization scheme has been proposed, which is proven to minimize the outage probability on the premise of guaranteeing fixed detection performance. Theoretical analysis and simulation result show that the optimal scheme can improve outage performance of the cognitive system and enhance transmission reliability of the cognitive network.
A multi-mode GNSS signal simulator, which is capable of simulating GPS L1, GLONASS L1, Beidou1, Beidou2 B1 and B3 signals, has been developed based on the NI PXI RF Test Hardware Platform. Technologies such as satellite navigation, software radio have been applied in the development process. DDS-based IF signal synthesis and three-stage up-converting methods have been chosen for simulated signal generation. Its software composition and execution efficiency optimization by hybrid programming and multi-threading are especially discussed. Simulated signals have been verified by off-the-shelf GNSS receivers.
An interferometric-like APES algorithm is proposed for blind DOA estimation of interferometric arrays. The number of targets and coarse direction cosine estimates are obtained from the interferometric-like APES spatial spectrum and model-order selection criterion. The fine direction cosine estimates are derived from the phase center's shift of subarrays. A dual-size algorithm is used to resolve the ambiguity in DOA estimation, and then high accuracy and unambiguous DOA estimates are achieved. The proposed approach is a blind DOA estimation method with higher accuracy than MUSIC and dual-size ESPRIT algorithms. Simulation results and real data processing demonstrate high accuracy of DOA estimation of interferometric arrays and the validity and feasibility of the proposed method, which can provide reference for the design of the interferometric array.
With the rank-1 constraint, a method for camera intrinsic parameters calculation based on spheres is proposed. The geometric interpretation of the rank-1 constraint is explored from the relation of the image of the absolute conic and the sphere image. Three algebraic equations are derived from the rank-1 constraint. The relation of the sphere images and the camera intrinsic parameters is clearly interpreted in vision geometry. Finally, a nonlinear optimized method is used with the three constraint equations for improving the accuracy. Experimental results show that our method is robust. Compared with the traditional method, it can improve the accuracy of the results of the intrinsic parameters.
We focus our attention on how to get optimal performance of concurrent transmission in heterogeneous networks. An equivalent queuing model is used to analyze the end-to-end delay performance of perfect splitting strategy with the theoretical delay bound found. Furthermore, two splitting algorithms, Minimum Queuing Delay based splitting strategy (MQD) and Unified Queuing Management Based Splitting Strategy (UQM), are presented based on the joint resources scheduled in different RATs (Radio Access Technology). In the two-RAT scenario, the two-dimensional discrete-state continuous-time Markov process is used to analyze our schemes with close-formed solutions found. Simulation results demonstrate that our proposed flow splitting strategies utilize the system resources efficiently and outperform current strategies, getting close to the theoretical delay bound especially in the heavy load region.
A rate-compatible puncturing algorithm for network LDPC codes is proposed to achieve the minimum error probability, and decoding error probability is derived. The algorithm for finding variable nodes to be deleted based on the tanner graph of network LDPC codes is proposed, which is aimed at minimizing the decoding error probability and optimizing the puncturing pattern. Simulation results illustrate that, at BER=10-4, the proposed rate-compatible network LDPC codes have a coding gain of about 0.4dB compared with the codes obtained via the existing punctured methods.
In order to improve the accuracy of aurora images classification, an algorithm based on the wavelet hierarchical model is proposed. In the proposed algorithm, the global and local wavelet features are extracted hierarchically first, then reduced in dimensions through the principal component analysis and used to classify the arc and three corona aurora images by the use of the support vector machine. By comparing the classification accuracy and time consumption, the optimal parameters in the wavelet hierarchical model are experimentally obtained and the validity of principal component analysis in feature optimization is verified. Experimental results show that the proposed algorithm improves the classification accuracy to a great degree with an acceptable time consumption compared with classical algorithms. Classification results between each two types of aurora images also provide some potential ways to improve the accuracy.
