最新刊期

This paper introduces a multi-component time-frequency analysis method, which conducts matching demodulation transform with a Kalman filter. This method builds the model on the local frequency properties of the signal, with model parameters seen as the state variables and the TF ridges treated as the observations. With the state and observation equation established, Kalman filter can be used to estimate the frequency of the signal. Finally, demodulation transform is used to generate a higher energy concentrated time-frequency distribution. The simulation results show our method has a better performance than existing methods under various SNR circumstances.   

Wireless network-on-chip (WiNoC) has emerged as a promising paradigm for the multi-core interconnection. However, the congestion problem can severely degrade network performance. There are two kinds of congestion in wireless network, node-level congestion and link-level congestion. To alleviate those problems, a congestion-avoidance wireless router is proposed in this paper. For the node-level congestion, we design a scheme that allows wireless routers to be able to acquire congestion information of each other through broadcasting information in wireless channel. Thereby, wireless router with high congestion degree can be avoided by overladen data injection. For the link-level congestion, we design a parallel FIFO scheme that allows transmitting three data flits within a single cycle. In addition, a mathematical modeling is established to analyze the performance of parallel FIFO scheme. Experiments show that our work has a better performance in reducing the wireless network congestion and improving wireless resource utilization rate compared to the other WiNoCs.   

In order to achieve the high QoS requirements demanded by many users in the same geographic area in a finite bandwidth case, the performance of providing packet-level QoS in the broadband wireless access network based on the IEEE 802. 16 standard is studied in this paper. In practice, firstly, the arrival process and the traffic source is modelled by Batch Markov Arrival Process and Continuous Time Markov Chain and the more practical and more accurate queuing model is achieved. Then, based on the state space, a wireless access network system is modeled. By the in-depth analysis for the obtained system model and the queuing model achieved previously, the QoS performance metrics about this model, such as the average queue length, packet dropping probability, the queue throughput and the average packet delay, are obtained. The simulation results show that the proposed analytical model in this paper can improve and enhance the QoS performance metrics compared with other typical analytical models.   

To construct a tracking-resistant P2P botnet, a Cross-Realm Architecture (CRA) was proposed. CRA strictly restricts bots' interactions across different realms and hides the origins of bots' interactions by IP spoofing. Considering the infeasibility of monitoring the global Internet and the difficulty of IP traceback, it is very hard for defenders to track CRA in the real world. The simulation results show that compared to recent popular P2P botnet architectures, CRA has better anti-tracking performance and robustness.   

Random measurement matrix needs large storage space, huge memory requirements for reconstruction, and high computational cost, which are not suitable for large-scale applications. To reduce the storage space of random measurement matrix for compressed sensing (CS), a new sampling approach for CS with semi-tensor product (STP-CS) is proposed. The STP-CS approach generates a random matrix, where the row and column numbers of the matrix are smaller than that for conventional CS. Then we optimize the matrix by the singular value decomposition (SVD) approach, after sampling with the matrix, we estimate the solutions of the sparse vector with the smooth l₀-norm minimization algorithm. Numerical experiments were conducted using gray-scale images, the peak signal-to-noise ratio (PSNR) and the structural similarity index (SSIM) of the reconstruction images were compared with the random matrices with different dimensions. Comparisons were also conducted with other random measurement matrix and other low storage techniques. Numerical experiment results show that the STP-CS can effectively reduce the storage space of the random measurement matrix to 1/256 of that for conventional CS, while maintaining the reconstruction performance.   

The classification of the wheeled vehicles and tracked vehicles is one of the difficulties of vehicles recognition. Rotation of the wheels and running of the track are the typical forms of micro-motion, the micro-Doppler signatures of the micro-motion can be the crucial proof for classification of the two kinds of vehicles. Firstly, the micro-Doppler difference of different micro-motion of the two vehicles are analyzed based on the echoes under radar with short dwell time;Secondly, based on the short-time stability and correlation of the echoes, the Auto Regression (AR) model of echoes is established, the forward and backward prediction method is applied in the extraction of AR model coefficients, which are used in the classification of the two kinds of vehicles, and the selection of the order of AR model is operated, the separation abilities between forward and backward prediction coefficients and forward prediction coefficients are compared;At last, after the pre-processing of the measured data, the classification of the two vehicles is accomplished with high accuracy based on the extraction of AR model coefficients. The effectiveness and robustness of the method are proved by the simulation results.   

For a class of unknown non-uniformly sampled discrete time nonlinear non-affine systems, the corresponding model free adaptive controller is designed by using equivalent data model in the current working point and the non-linear estimation of pseudo Jacobi matrix (PJM) by using input and output data. The proposed control method has the following characteristics:the designed controller requires only the non-uniformly sampled input and output data, and does not contain any information of the system model. The control algorithm has small calculation amount, and is easy to achieve. Finally, the effectiveness of the proposed method is verified by the simulation results of a non-uniformly sampled discrete-time nonlinear system.   

In order to improve the unbalancing problem of battery caused by the difference between single cells during the charging and the discharging processes, an improved energy storage circuit and a new type of battery management system were designed based on the study of flying-capacitor and charging power balance method, while at the same time, a balance control method was developed. A test was carried out on the wireless receiver of gastrointestinal robot, showing that the battery management system can work on the type of assisted charging balance as well as the type of energy consumption balance, and has more than twice the actual capacity of the battery pack and its utilization, compared to the previous balanced control method.   

A new level set method has been proposed for coastline detection in high-resolution SAR images based on the generalized Gamma distribution (G Γ D). The G Γ D is a statistical model with high flexibility, which is able to characterize the diversity of scenes in SAR images effectively. The parameter estimation of the G Γ D is realized by the method of log-cumulants. Then the energy functional is formulated based on the G Γ D. The coastline detection is achieved by minimizing the proposed energy functional using the level set segmentation method. Experimental results with measured TerraSAR-X images have demonstrated that the proposed method can obtain more precise coastline detection results.   

In order to segment the helicopter target in the infrared image with complex cloud background effectively, a variational segmentation model based on manifold feature and shape priori is proposed. The region energy term is constructed on the basis of gray distribution of the image data to drive the deformation curve to target boundary. A regional covariance descriptor is introduced on Symmetric Positive Definite (SPD) matrix manifold to define a prior feature energy term to improve the robustness of the segmentation algorithm. The segmentation result of the helicopter in the first stage of the infrared image is obtained by the combination of the region term, the boundary term and the prior feature term. In the second stage, Principal Component Analysis(PCA) is used to capture the prior shape variation pattern of the infrared helicopter target, and the shape of evolution curve is reconstructed in PCA space as the prior knowledge to restrain the curve deforming. Finally, the complete contour of the infrared helicopter is obtained. Experimental results demonstrate that this method can effectively extract the whole contour of the infrared target.   

In B-ISAR imaging, the range resolution and cross-range resolution are dependent on the signal band and the coherent processing interval, respectively. Generally, the B-ISAR image is seriously affected by noise. In this paper, a matrix form of complex approximate message passing algorithm based on two dimensional coupled dictionaries (MCAMP-TCD) is presented, by considering the 2D coupling sparse feature of the echo. Firstly, the range-azimuth 2D joint B-ISAR imaging model is established. Then the 2D joint super-resolution imaging problem is converted into a complex basis pursuit denoising (C-BPDN) problem through vectorization operation. Secondly, two strategies are implemented to solve C-BPDN problem quickly, the first strategy is utilizing the relation between vectorization operation and Kronecker product to derivate the matrix form of complex approximate message passing algorithm, which can avoid the high computational complexity and memory requirements due to vectorization operation. In second strategy, the two dimensional fast Fourier transform (2D FFT) is introduced to equivalent matrix multiplication, which further reduces the computational complexity of the single iteration. At last, the imaging capability under low signal to noise ratio (SNR) is improved by the ability to accurately approximate the noise threshold of the MCAMP-TCD. Simulation results verify the effectiveness of the proposed method.   

Content sharing plays the essential role to achieve the vision of "The Internet is for Everyone". To this end, Prof. Li Youping, academician of the Chinese Academy of Engineering, pioneered the concept of DAN (Dual-Architecture Network), which is composed of primary architecture (refers to Internet) and secondary architecture (refers to broadcast-storage network). In this paper, firstly the implementation model of DAN and its formalized description are exhaustively presented, and then qualitative analyses for the content sharing mechanism and content business model are conducted. Furthermore, quantitative analysis is conducted by applying the 2ACT model. Finally, the feasibility is verified through a prototype system. The content sharing capability of DAN is eventually demonstrated in terms of theoretical analysis and system implementation.   

With the development of the simulation technology of the magnetic field, there are several kinds of magnetic sources that can simulate the target magnetic field with a high degree of accuracy, but it's difficult to measure its similarity in traditional ways. To solve this problem, a new similarity measurement eigenvalue of magnetic field based on discrete Fréchet distance is proposed in this paper. To get the eigenvalue, the evaluation index of the distance between discrete points in Fréchet method has improved according to the characteristic of magnetic field curve, after that, a parameter which can measure the degree of similarity of the horizontal coordinate and eliminate the different speed of the target is proposed, then the similarity measurement eigenvalue is obtained, at last, its feasibility is verified by examples. Compared with the traditional methods, this eigenvalue can be used to evaluate the similarity of the closer magnetic field curves.   

Detecting difference between process models is one of the key technologies in business process management. The classical graph edit distance cannot directly be used to detect the difference between process models because there are many kinds of nodes in a process model. To solve this problem, we present an algorithm for detecting difference between process models based on process structure tree. Firstly, the process models are converted to their corresponding task based process structure trees (TPSTs). Then the divide and conquer strategy is used to obtain the optimal mapped node set between two TPSTs. Finally, an edit script with approximate minimum cost is generated based on the mapped nodes, which is considered as the difference between two process models. The experimental results show that this algorithm can meet the real requirements in terms of precision and efficiency.   

For traditional algorithms, activities are treated as events that are independent from each other and cannot be accurately identified, we propose a workflow task recognition algorithm based on intelligent programming, which utilizes the commonality of workflow and intelligent planning in the implementation of sequence and operation rules. Through the high-level abstract description, we can deduce the intrinsic logic relation between them automatically, and also explore potential knowledge from outside information, and transform the workflow task recognition problem into the corresponding planning recognition problem, and solve the problem that the traditional algorithm is difficult in recognition and sensitive to the noise data. Experimental results show that our algorithm is feasible.   

The long-term information of speech signals shows excellent performances in the applications of voice activity detection. Six types of long-term information based on auditory filter banks are proposed through the non-linear spectral decomposition with three different auditory filters. Further, an adaptive voice activity detection algorithm based on these types of long-term information is proposed. Without additional training data, this algorithm use the data selecting from the test signals according to long-term information to train a speech/non-speech classifier, and classifies the current test signals using the speech/non-speech classifier frame by frame. Experiments on TIMIT dataset and NOISEX-92 dataset show that the algorithm improves the performance of VAD with higher accuracy and stronger robustness in low SNR noisy environments. The online experiments show that it can also obtain a good performance in real-time processing conditions.   

In radio frequency identification (RFID)systems, excepting to identify the basic information of target, how to additionally aware the precision position of the target is an important and pressing research topic. In this paper, we propose a novel range free tag localization scheme based on phased array antenna for RFID system, called PATL. This method takes advantage of the adjustable radiation angle of the phased array antenna to scan the surveillance region in turns. By using the statistics of the tags' number in different antenna beam directions, a weighting algorithm is used to calculate the position of the tag. This method can be applied to real-time location of multiple targets without usage of any reference tags or additional readers. We use a COTS UHF RFID reader which is integrated with a phased array antenna to evaluate our method. The results show the average is about 21cm and the time consumption also satisfy the real time performance.   

Clustering can increase the performance of large-scale mobile ad hoc networks effectively. But the highly dynamic mobile ad hoc networks have some of characteristics, such as high mobility and fast network topology change;applying traditional clustering algorithms will cause the sharp decrease of the performance of the network, frequent updates of cluster topology will cause the instability of cluster structure and the increase of control overhead. For purpose of solving the problems that traditional clustering algorithms cannot fit the highly dynamic large-scale mobile ad hoc networks, BIMAC (Bio-Inspired Mobility-Aware Clustering) algorithm is proposed. This algorithm ameliorates the forage model of physarum polycephalum, which can make it adapt to the domain of mobile ad hoc networks. On account of this algorithm includes the mobility characteristic of the node, we can carry through the cluster formation and maintenance effectively. Experimental results have indicated that the BIMAC algorithm increases average link connection lifetime and average cluster head lifetime compared with other traditional clustering algorithms;BIMAC algorithm can make the cluster structure more stable. This algorithm can increase the adaptive capacity for highly dynamic large-scale mobile ad hoc networks.   

Local multiple kernel learning method could learn a specific combination kernel function for various samples according to the local space characteristics, therefore it has better discriminant ability. In this paper, we propose a local variable sparsity based multiple kernel learning method. In our method, the samples are divided into a few groups with a soft grouping method and the sparsity of kernel weights in various local spaces is determined by the similarity of kernels. We use an alternative optimization method to solve this problem. The experiment on synthetic dataset indicates that our method has a strong advantage in discriminative feature learning and against noise. Finally we apply our method into image scene classification and the accuracy is improved obviously.   

A fully homomorphic encrgption scheme was presented based on number theory research unit (NTRU), which is provable security about indistinguishable chosen plaintext attack (IND-CPA). Firstly, to avoid the effective lattice attacks, we modified the key generation algorithm of NTRU by Gaussian abstraction algorithm of lattices, and the distribution of the key is not changed. Then, we proposed a new homomorphic encryption scheme based on the improved NTRU encryption algorithm by using the Flattening technique. Its IND-CPA security was proved strictly under the standard model.   

This paper constructed a calibration system of oscilloscope based on the optical pulse source. The 70GHz sampling module 86118A of Agilent 86100C was calibrated based on optical pulse source using the standard pulse method. In the calibration process, the influences on calibration results of three kinds of time base errors were researched for the first time, e. g. , time base drift was removed using a cross-correlation algorithm, time base distortion was corrected using the least squares method, and jitter was removed by using PDF deconvolution. Experiment results showed that calibration of 70GHz broadband oscilloscope was realized, the time base errors could be effectively removed by this method, and the accuracy of the calibration was satisfied.   

A class of planar bandpass filters (BPFs) using dual/triple-mode substrate integrated waveguide (SIW) resonator and ground coplanar waveguides (GCPWs) are proposed according to miniaturization and high selectivity. The quasi-elliptic, asymmetric and non-transmission zero (TZ) response dual-mode SIW filters can be realized by changing resonance frequencies of TE₁₀₂ and TE₂₀₁, and the external couplings. Two GCPWs of the same size are etched on SIW cavity top metal plane;and the TE₁₀₂, TE₂₀₁ and two GCPWs in a single SIW cavity are combined to construct the diversified response fourth-order filters. On the basis of asymmetric and non-TZ response, the fundamental mode TE₁₀₁ can be shifted to the passband to form a fifth-order filter with wider bandwidth. The designed BPFs were fabricated and measured;and the simulated and measured results agree with each other, demonstrating the feasibility of using hybrid dual/triple-mode SIW and CPWs in a single cavity to design BPFs.   

The arithmetic cell, especially the multiplier, is one of the key components of the modern digital systems, which could affect the efficiency of the whole system dramatically. In this article, we propose an extended Booth algorithm, which is different from the traditional one with the strategy of keeping all the partial products during shifting, then compressing and finally summing. This new one could reduce the coupling of each function and simplify the control part of the circuit. Moreover, by following the asynchronous design methodology, the control circuit with the Click based asynchronous micro pipeline that obeys a two-phase handshake protocol with "bounded bundle data" is implemented. With the policy of isolating data processing from control, an 8-bit asynchronous multiplier is designed and then implemented with FPGA. With the FPGA of 45nm technology and the same architecture, this asynchronous Booth multiplier is over 12 times faster than the synchronous one while keeping the area and power almost same.   

To improve the compression performance, a frame memory lossless compression using grain-directional adaptive prediction and Golomb run-length entropy coding is proposed. First, the optimal reference pixels were obtained through double-scan and multi-directional prediction;then, the pixel differences were calculated and entropy-coded through Golomb and run-length. In this way, the compression performance is improved. Simulation results show that compared with those of reference frame lossless compression based on intra prediction and Golomb entropy coding, the average compression ratio is increased by 16% while the average coding time is reduced when using the proposed algorithm.   

In this paper, the problem of robust stability analysis for a class of linear uncertain systems with interval time-varying delay is studied. In order to develop a less conservative stability condition, a Lyapunov-Krasovskii functional comprising quadruple-integral term is introduced. A novel delay dependent stability criterion in terms of linear matrix inequalities (LMIs) is given by using delay-central-point (DCP) method and reciprocally convex combination technique, which is derived by integral inequality method. Compared with the existing literature, this criterion can greatly reduce the complexity of theoretical derivation and computation. Finally, three well-known numerical comparative examples are given to verify the superiority of the proposed approach in reducing the conservation of conclusion.   

In the case of the delay and error Channel State Information (CSI), a robust joint interference and phase alignment algorithm based on Bayes estimation and power allocation among signal flows is proposed for multi-cell MIMO-BC. Firstly, the best prediction of the current CSI is obtained through Bayes estimation by the senders. Secondly, interference suppression matrix is designed through maximizing signal to inter-cell interference plus noise ratio in the forward link, and the pre-coding matrix is designed through maximizing SINR in the reverse link. Further, the water-filling power allocation is used to optimize power allocation among signal flows. Finally, the interference among the data flows is rotated into signal space which improves the received power. Under the conditions of the perfect CSI and the delay and error CSI, the simulation results verify that the proposed algorithm improves the performance of system compared with the existing algorithms.   

With the increasing deployments in IP Cores in FPGA applications, SoC-FPGA systems have been gaining wide popularity in recent years. Processor is no doubt the heart of a SoC-FPGA system, and has a critical impact on overall system performance. The utilization of open source processor IPs can greatly improve working efficiency and reduce the development cycle of SoC-FPGA systems, and it has been gaining favor as a commonly adopted design methodology. This paper investigates the key technical indicators of existing open-core projects, and provides an overview of open source processors. The major features of open source processors are summarized in terms of usability and stability, and the authors propose a methodology to choose appropriate processors for specific applications. Following these guidelines, some open source processors are selected and implemented on Stratix V and Virtex-7 FPGA platforms using corresponding EDA tools. The implementation results are compared and discussed.   

For the modeling issue of non-uniformly multi-rates sampled nonlinear systems, a fuzzy identification method based on hierarchical principle is proposed in the paper. First of all, a nonlinear system is described as a weighted combination representation of the multiple local linear models by using lift technology when the non-uniformly updating scheme for input signals and uniformly sampling scheme for output signals are taken in the data sampling process. On this basis, we propose a fuzzy identification algorithm, in which the GK fuzzy clustering method and a recursive least squared method based on hierarchical principle are used to confirm the premise structure and consequence parameters of fuzzy model, respectively. Moreover, the convergence of the identification algorithm is studied by using martingale theorem. Finally, the effectiveness of the proposed method is demonstrated by a simulation example.   

In order to solve the problem that how to explore and exploit the high-order label correlations effectively in multi-label learning, a Multi-Label evolutionary HyperNetwork based on label Correlations (MLHNC) is proposed in this paper. In MLHNC, the predicting results obtained from any multi-label learning method are utilized as input of the model, the high-order correlations among labels are represented and explored by hyperedges, and the final prediction is made by integrating the label correlation and feature information. The experimental results on six multi-label datasets compared with three state-of-the-art multi-label learning methods show that the MLHNC not only improves the performance of various state-of-the-art multi-label learning methods, but also provides readable learning results.   

In this paper, we propose an identity-based ring signature scheme based on Waters dual system encryption technology and the orthogonality property of composite order bilinear group operation. The scheme, relying on two simple static assumptions, is fully secure in the standard model. Due to the merit of Hierarchical identity-based encryption (HIBE), the proposed ring signature scheme achieves unconditional anonymity and has much higher computational efficiency.