Search

A compact microwave bandpass filter with λ/4 stepped impedance resonators (SIRs) is presented in this paper. λ/4 SIRs are employed here not only to miniaturize the circuit size, but also to eliminate spurious resonance at 2f 0 . To achieve the maximal number of finite transmission zeros, this filter is synthesized based on the 4th-order fully canonical coupling topology. As a result, four finite transmission zeroes are analytically achieved, and used to sharpen filter rejection skirts. Three capacitive J couplings are systematically designed to meet the required coupling strengths for the filter frequency response. Besides, they also adjust their coupling zeros for 3f 0 spurious suppression. Lastly, a 4th-order fully canonical bandpass filter is designed and fabricated with central frequency of 1.8 GHz and fractional bandwidth of 11.0% The proposed filter possesses preferred merits: compact size, 4 transmission zeros and up to 4.8f 0 20 dB attenuated stopband.

Based on the Location Quotient (LQ) method, this paper selected datas of Statistical Yearbook from 2001 to 2008, and calculated the LQ of all sectors of industry in Xinjiang, finding three industries has the phenomenon of clustering, namely Extraction of Petroleum and Natural Gas industry, Oil Processing, Coking and Nuclear Fuel Processing industry and Processing of Ferrous Metals Ores industry. The cluster degree of all sectors has also been calculated, sorted and analyzed. In addition, changes of cluster degree from 2001 to 2008 in clustered industrial sectors was described and reasons for, and some suggestions on the industrial cluster development has been put forward of this region.

In this paper, we focus on the identification of discrete-time piecewise affine (PWA) systems from noisy data. This problem consists of the estimation of both the local affine subsystems and the partition of the regression space. A two-stage robust identification approach is proposed to estimate the local affine subsystems in the presence of noise. This approach includes an optimization-based initial estimation process and a least-squares-based refinement procedure. In addition, to estimate the partition of the regression space for continuous dynamic PWA systems, an intersection approach is proposed as an alternative to the widely used pattern recognition approaches. Simulation studies demonstrate the effectiveness of the two-stage identification approach and the intersection approach in noisy case.

The aim of this paper is to introduce a novel phase-based feature representation for robust speech recognition. This method consists of four main parts: autoregressive (AR) model extraction, group delay function (GDF) computation, compression, and scale information augmentation. Coupling GDF with an AR model results in a high-resolution estimate of the power spectrum with low frequency leakage. The compression step includes two stages similar to MFCC without taking a logarithm of the output energies. The fourth part augments the phase-based feature vector with scale information which is based on the Hilbert transform relations and complements the phase spectrum information. In the presence of additive and convolutional noises, the proposed method has led to 15% and 12% reductions in the averaged error rates, respectively (SNR ranging from 0 to 20 dB), compared to the standard MFCCs.

The paper studies the robust control problem of discrete-time systems with state saturation. First the paper introduces an ellipsoid and supposes that the invariant set of the system contains the ellipsoid. By using matrix norm theory we obtain one sufficient condition of the stability of the discrete-time system with state saturation. The attraction domain of the system is estimated. Then the paper researches the stability of the uncertain system with state saturation. Finally a numerical example is presented to demonstrate the result of the paper.

This paper considers the characteristic mode analysis (CMA) of arbitrary electromagnetic structures using the comprehensive 3D electromagnetic field solver, FEKO [1]. The theory of characteristic modes, as presented in [2], is used to derive the real orthogonal current modes. These modes are obtained by solving a generalised symmetric eigenvalue problem defined by the real and imaginary parts of the Method-of-Moments (MoM) impedance matrix. The research presented in this article discusses the techniques used in FEKO to solve this generalised eigenproblem. Furthermore, paralleli-sation using both distributed and shared memory programming models, as well as GPU computation is considered within the FEKO framework to accelerate the CMA.

Particle swarm optimization (PSO) is a nature-inspired technique for solving continuous optimization problems. For a fixed optimization problem, the quality of the found solution depends significantly on the choice of the algorithmic PSO parameters such as the inertia weight and the acceleration coefficients. It is a challenging task to choose appropriate values for these parameters by hand or mathematically. In this paper, a novel self-optimizing particle swarm optimizer with multiple adaptation layers is introduced. In the new algorithm, adaptation takes place on both particle and subswarm level. The new idea of using virtual parameter swarms which hold modifiable parameter configurations each is introduced. The algorithmic PSO parameters can be mutated by using, for instance, well-known techniques from the field of evolutionary algorithms, in order to allow fine-granular parameter adaptation to the problem at hand. The new algorithm is experimentally evaluated, and compared to a standard PSO and the Tribes algorithm. The experimental study shows that our new algorithm is highly competitive to previously suggested approaches.

This paper deals with 3D shape reconstruction based on a structured-light projector and two cameras using Evolutionary Algorithms (EAs). Two main problems are solved: first, the correspondence problem (also called the stereo matching problem) which allows the matching of different views of a same point in the two images, and second, the camera calibration which takes into account the camera parameters and the camera relative locations through a fundamental matrix. Concerning the correspondence problem, the main issue is the detection of the points to be matched, i.e. the points of interest (POIs). In this paper we propose a structured pattern to be projected on the object to analyse. The structured pattern is composed of triangles with different colors and different orientations. The POIs to be matched are the vertices of the triangles. Then we define an encoding method allowing the recovery of all the POIs that are detected in both images. In previous work, we showed that EAs can be used to obtain the fundamental matrix, thus providing an uncalibrated 3D reconstruction approach. However the method processed too small a number of POIs. The proposed acquisition system allows a high number of POIs to be obtained. The initial EAs are improved to use the new set of POIs. The experiments show that the proposed encoding method allows an efficient matching. In addition, accurate 3D reconstruction results are obtained and the process converges correctly.

This paper presents an idea to further simplify and relax the linear matrix inequality (LMI) stability conditions of Takagi-Sugeno (T-S) fuzzy systems. By considering the distribution of membership functions in a unified space, we can easily find the overall relation of the original nonlinear system and its approximated local subsystems. Based on the theory of convex combination, the upper-bounds and lower-bounds of each membership functions can be directly used to construct less conservative LMI conditions. The cases of both Type-1 and Type-2 systems are considered, and examples are provided to illustrate the achieved improvement.

Stack filters are a class of nonlinear digital filters possessing threshold decomposition and stacking property. In this paper, new nonlinear threshold decomposition architecture is introduced to reduce the complexity of algorithm. Then by combining with the character of image, a dynamic nonlinear decomposition architecture is proposed to implement stack filters which can shorten the running time in the filtering procedure and improve the quality of the output image simultaneously. Comparisons on the mean absolute error (MAE), mean square error (MSE), power signal-to-noise ratio (PSNR) and on the time of filtering computation between this new algorithm and others are provided to show the validity of proposed algorithm.

In this paper, three main criteria for the adaptive beamformer design are discussed: maximal signal-to-interference-plus-noise ratio (MSINR), minimal mean-squared error (MMSE), and minimal least-square error (MLSE). Although in the case of exactly known power and steering vector of the signal-of-interest (SOI), there are beamformers that can simultaneously meet the MMSE and MSINR criteria, this is no longer true when the exact knowledge of the steering vector is unavailable. To account for steering vector errors, a meaningful approach is to model the actual steering vector as random. In this paper, it is shown that in the latter case, the MMSE and MSINR criteria can not be simultaneously attained. We study the achievable region in the MSE-SINR plane and propose a new adaptive beamformer that can attain a frontier of operating points on the boundary of this region and, therefore, provide an optimal performance tradeoff among the MSINR and MMSE criteria. It is also shown that in the random steering vector case, the MLSE and MSINR criteria are simultaneously achievable and a new adaptive beamformer is proposed that satisfies both these criteria.

Discriminant Analysis is utilised in working out which specific classification, a data pertains to on the basis of its needed features. Linear Discriminant Analysis(LDA) achieves the maximum class separability by projecting high-dimensional data onto a lower dimensional space. However, LDA suffers from small sample size(SSS) problem where the dimensionality of feature vector is very large compared to the number of available training samples. Regularized Discriminant Analysis(RDA) handles SSS problem of LDA with an introduction of regularization parameter(λ) and has the ability to reduce the variance. One important issue of RDA is how to automatically estimate an appropriate regularization parameter. In this paper, we propose a new algorithm to enhance the performance of RDA by effectively estimating an appropriate regularization parameter in order to reduce training time and error rate. Experiments are done using various benchmark datasets to verify the effectiveness of our proposed method with the state-of-the-art-algorithm.

This paper presents an analog implementation of the wavelet transform (WT) in an ultra low power environment, using the dynamic translinear (DTL) circuit techniques. The circuit consists of an analog filter whose impulse response is the first derivative of a Gaussian. A convenient method to provide the transfer function of the filter is given by the Pade approximation. In order to fulfil the low-power requirement, the filter's state space description is optimized with respect to dynamic range. Finally, the DTL principle is applied to implement the state space filter. Simulations indicate a good approximation of the wavelet transform.

Chaotic maps are an alternative for calculating pseudorandom numbers which have created an increased interest among researchers dealing with stochastic search and optimization algorithms in the recent past. This interest is based on promising results with respect to both the quality of the results as well as the running time of the optimization algorithms compared to the usually used standard pseudorandom number generators. In this paper we investigate the influence of nine different chaotic maps on the quality of the results obtained by a self-organizing map (SOM) which has been used to solve the Traveling Salesman Problem (TSP). The investigation is based on various sizes of both the problem instances as well as the number of iterations where all nine chaotic maps are compared against the pseudorandom number generation. As a result it is proven that chaotic maps are significantly better in several cases. Finally, possible reasons for both the superiority and inferiority of chaotic maps compared to pseudorandom number generation are analyzed and discussed.

This paper considers the problem of sequential assignment of probabilities (likelihoods) to elements of an individual sequence using an exponential family of probability distributions. We draw upon recent work on online convex programming to devise an algorithm that does not require computing posterior distributions given all current observations, involves simple primal-dual parameter updates, and achieves minimax per-round regret against slowly varying product distributions with marginals drawn from the same exponential family. We validate the theory on synthetic data drawn from a time-varying distribution over binary vectors of high dimensionality.

In this paper we obtain a Maximum Likelihood Estimator (MLE) for average received power in a multiuser non coherent multiple frequency-shift keying (MFSK) system over a fast Rayleigh fading channel. It is shown that this estimator is unbiased and that its variance can be obtained by a simple linear function. The Cramer Rao lower bound (CRLB) is derived and compared to the MLE's variance. It is shown that in most of the practical cases the MLE is inefficient by a multiplier that depends on system parameters but not on the number of sample points used. Numerical results are obtained and average performance is analyzed.

Several optimal design schemes of a single-sided linear induction motor (SLIM) adopted in linear metro are presented in this paper. Firstly, the equivalent circuit of SLIM fully considering the end effects, half-filled slots, back-iron saturation and skin effect is proposed, based on one-dimensional air-gap magnetic equations. In the circuit, several coefficients including longitudinal end effect coefficients K r (s) and K x (s), transversal end edge effect coefficients C r (s) and Cx(s), and skin effect coefficient K f are achieved by using the dummy electric potential method and complex power equivalence between primary and secondary sides. Furthermore, several optimal design restraint equations of SLIM are provided in order to improve the operational efficiency and reduce the primary weight. These nonlinear equations are solved by using genetic algorithm and mixture penalty function method. The optimal schemes are compared with the original design of one company, where analysis on parameters is made in detail. These results show that the optimal schemes are reasonable for improving the performance of SLIM.

This paper investigates the stability of event-based state-feedback loops. Two analysis methods are proposed. The first method proves the event-based state-feedback loop to be input-to-state practically stable which means that the state converges to a vicinity of the origin. The size of this region depends upon the magnitude of the disturbance and the event threshold, which is a design parameter. The second method is tailored for event-based state-feedback loops with stable plant dynamics and it shows that the investigated system is input-to-state stable, which implies that for small disturbance magnitudes the size of the region to which the state converges is independent of the event threshold. This new result shows that asymptotic stability can be achieved by means of an event-based controller with constant event threshold which has been proven in literature only for decreasing event thresholds. Both analysis methods are applied to a benchmark example and the results are compared with an analysis method known from literature which shows that the second proposed analysis method yields less conservative results with respect to the ultimate bound than existing methods in literature.

Diffusion weighted imaging (DWI) is a non-invasive method for investigating the brain white matter structure. It can be used to evaluate fiber bundles in the brain. However, clinical acquisitions are often low resolution. This paper proposes a method for improving the resolution using sparse representation. In this method a non-diffusion weighted image (bO) is utilized to learn the patches and then diffusion weighted images are reconstructed based on the trained dictionary. Our method is compared with bilinear, nearest neighbor and bicubic interpolation methods. The proposed method shows improvement in terms of Peak Signal-to-Noise Ratio (PSNR) and Structural SIMilarity (SSIM).

In this paper, we propose an efficient real time tracker that uses a differential evolution strategy within the particle filter framework. Particles are strategically propagated based on the maximum a posterior (most likely) object location with genetic operators. This enables the use of a small sample size and alleviates the frequent sample degeneracy and impoverishment problems encountered in particle filters. We reduce the sample size considerable while improving the trackers accuracy. This makes the proposed tracker a good candidate for real time object tracking or an embedded resource constrained tracker.

This paper describes a nonlinear fuzzy controller which can control highly nonlinear processes. In general nonlinear processes can be seen as linear processes working in a confined operating region, this region can be characterized by the input, the output of some of the state variables. The controller structure has two layers: the first layer is a fuzzy controller, representing the knowledge about the use of proportional and integrative controller, and the second one is a set of PI controllers. Some guidelines about how to tune this controller are also described. Simulations and experimental results show the potentialities and limitations of this approach.

In broadcast fading channel, channel variations can be exploited through what is referred to as multi-user diversity and opportunistic scheduling for improving system performance. To achieve the gains promised by this kind of diversity, the transmitter has to accurately track the channel variations of the various receivers, which consumes resources (time, energy, bandwidth), and thus reduces the resources remaining for effective data transmissions. The transmitter may decide not to acquire or probe the channel conditions of certain receivers, either because these receivers are presumably experiencing severe fading, or because the transmitter wishes to spare resources for data transmissions. It may also decide to transmit to a receiver without probing its channel; in such cases, the transmitter guesses the channel state, which often results in a reduction of the transmission rate compared to when the transmitter knows the channel state. Ultimately, the transmitter has to decide to which receiver it should transmit. In this paper, we identifying the joint probing and transmission strategies realizing the optimal trade-off between the channel state acquisition and the effective data transmission. The objective is to maximize the system throughput. Finally, we propose several extensions of the proposed strategy, including a scheme to maximize the system utility and a scheme to ensure the system stability., 10.1109/78.845940, © IEEE

In this paper a concept for a new method for a permanent, non-invasive blood pressure measurement in the ear is presented. Currently, blood pressure is measured about once a day and used as the basis for therapy. In order to enable an individual and as mild as possible therapy, a permanent, noninvasive measuring method is required. With every heartbeat, the arteries in the body expand and contract. Blood pressure is the pressure acting on an artery, the higher the pressure the greater the enlargement. If the external auditory canal is closed airtight, the increase and decrease in size of the arteries during the heartbeat causes a change in volume of the closed air chamber and thus a pressure fluctuation within it. The theory is, that these fluctuations represent the blood pressure. In order to prove the theory a prototype was set up, containing a pressure sensor (integrated into an Alpine InEar for airtight sealing), a micro controller and a reference ECG. The sensor was placed airtight in the ear. One test person showed perfect results, in which after each heartbeat in the reference system a corresponding signal was also visible in the ear system. The recorded curve resembles a blood pressure curve, which proves that it is in principle possible to measure blood pressure with such a system. For absolute blood pressure values, the system must be supplemented with further components, this is being researched in the project MikroBO11https://www.elektronikforschung.de/projekte/mikrobo funded by the Federal Ministry of Education and Research of Germany.

The Fisher Linear Discriminant (FLD) is commonly used in classification to find a subspace that maximally separates class patterns according to the Fisher Criterion. It was previously proven that a pre-whitening step can be used to truly optimize the Fisher Criterion. In this paper, we study the theoretical properties of the subspaces induced by this whitened FLD. Of the four subspaces induced, two are most important for classification and representation of patterns. We call these Identity Space and Variation Space. We show that only the between-class variation remains in Identity Space, and only the within-class variation remains in Variation Space. Both spaces can be used for decomposition and representation of class data. Moreover, we give sufficient conditions for these spaces to exist. Finally, we also run experiments to show how Identity and Variation Spaces may be used for classification and image synthesis.

Feature vector extraction, based on local image texture, is a primitive algorithm for many other applications, like segmentation, clustering and identification. If these feature vectors are a good match to the human visual system (HVS), we can expect to get the appropriate results by using them. Gabor filters have been used for this purpose successfully. In this paper we introduce a novel refinement, with the use of matching pursuits (MP) to improve the Gabor based texture feature extractor. With this improvement, we show that the separability of different textures increases. Another consideration in this work is computation complexity. Therefore, we limit the basis function set to reduce MP computation time.

In this paper, soft decision decoding of linear block codes based on the reprocessing of several information sets is considered. These information sets are chosen according to the reliability measures of the received symbols and constructed from the most reliable information set. Each information set is then reprocessed by the ordered statistics algorithm. General guidelines for the trade-offs between the number of information sets to be processed, the number of computations for reprocessing each information set, and the error performance to be achieved are provided.

In this paper, we propose a framework for isolating text regions from natural scene images. The main algorithm has two functions: it generates text region candidates, and it verifies of the label of the candidates (text or non-text). The text region candidates are generated through a modified K-means clustering algorithm, which references texture features, edge information and color information. The candidate labels are then verified in a global sense by the Markov Random Field model where collinearity weight is added as long as most texts are aligned. The proposed method achieves reasonable accuracy for text extraction from moderately difficult examples from the ICDAR 2003 database.

In this paper, we introduce a fractional differential HIV/AIDS infected model with Beddington-DeAngelis functional response rate. Moreover as for a feature, we consider cure rate of infected CD4 T cells. We show that the model introduced in this paper has nonnegative solutions which are all bounded. What is more, we also give a detailed analysis for the asymptotic stability of both the free equilibrium and the infected equilibrium. We have proven that if the basic reproduction number R 0 is less than unity, then the disease-free equilibrium is locally asymptotically stable. If R 0 is greater than unity, the infected equilibrium is locally asymptotically stable under some conditions.

Robust control is used for a system with uncertainties and disturbances. Above all, the H ∞ loop shaping method is known as the method achieving good control performance. However, this method takes much time to build the mathematical model of an actual plant and its uncertainty. In this paper, we propose the fixed-order controller design method which only uses frequency response data of the plant. Moreover, the controller is designed by the convex optimization. Thereby, the identification cost is expected to be less than the previous method without an increase in the calculation cost. The effectiveness of the proposed method is verified through the experiment for the belt-driven two-mass system, and comparison with the previous method in the last of this paper.

The 3D reconstruction of a face from a single frontal image is an ill-posed problem. This is further accentuated when the face image is captured under different poses and/or complex illumination conditions. In this paper, we aim to solve the shape recovery problem from a single facial image under these challenging conditions. The local image models for each patch of facial images and the local surface models for each patch of 3D shape are learned using a non-linear dimensionality reduction technique, and the correspondences between these local models are then learned by a manifold alignment method. By combining the local shapes, the global shape of a face can be reconstructed directly using a single least-square system of equations. We perform experiments on synthetic and real data, and validate the algorithm against the ground truth. Experimental results show that our method can yield accurate shape recovery from out-of-training samples with a variety of pose and illumination variations.

Data compression methods commonly focus on mean-square error (MSE) distortion. However, in TDOA/FDOA emitter location the distortion measure should focus on parameter accuracy. This paper discusses the proper choice of a distortion measure for this problem. We show that MSE is a poor choice for two reasons: (i) increased MSE may not even effect the parameter accuracy, and (ii) there are other factors that must be considered (e.g., the signal's RMS bandwidth/duration and the relative usefulness of TDOA vs. FDOA). We propose a new measure that captures the true impact of compression and illustrate its importance through a few simple examples.

It has been shown lately that any "standard" Bayesian lower bound (BLB) on the mean squared error (MSE) of the Weiss-Weinstein family (WWF) admits a "tighter" form which upper bounds the "standard" form. Applied to the Bayesian Cramer-Rao bound (BCRB), this result suggests to redefine the concept of efficient estimator relatively to the tighter form of the BCRB, an update supported by a noteworthy example. This paper lays the foundation to revisit some Bayesian estimation problems where the BCRB is not tight in the asymptotic region.

In this paper, the degrees of freedom (DoF) of the two-user single input single output (SISO) X channel are investigated. Three cases are considered for the availability of channel state information at the transmitters (CSIT); perfect, delayed, and no-CSIT. A new achievable scheme is proposed to elucidate the potency of interference creation-resurrection (IRC) when the available CSIT alternates between these three cases. For some patterns of alternating CSIT, the proposed scheme achieves 4=3 DoF, and hence, coincides with the information theoretic upper bound on the DoF of the X channel with perfect and instantaneous CSIT. The CSIT alternation patterns are investigated where the patterns that provide extraordinary synergistic gain and dissociative ones are identified.

Over the last years, particle filters (PF) have attracted considerable attention in the field of nonlinear state estimation due to their relaxation of the linear and Gaussian restrictions in the state space model. However, for some applications, PF are not adapted for a real-time implementation. In this paper we propose a new method, called box particle filter (BPF), for dynamic nonlinear state estimation, which is based on particle filters and interval frameworks and which is well adapted for real time applications. Interval framework will allow to explain regions with high likelihood by a small number of box particles instead of a large number of particles in the case of PF. Experiments on real data for global localization of a vehicle show the usefulness and the efficiency of the proposed approach.

Recent work by the author has extended the Supervisory Control Theory to include the class of control languages defined by pushdown machines. A pushdown machine is a finite state machine extended by an infinite stack memory. In this paper, we define a specific type of deterministic pushdown machine that is particularly useful as a discrete event controller. Checking controllability of pushdown machines requires computing the complement of the controller machine. We show that Exception Handling Controllers have the property that algorithms for taking their complements and determining their prefix closures are nearly identical to the algorithms available for finite state machines. Further, they exhibit an important property that makes checking for controllability extremely simple. Hence, they maintain the simplicity of the finite state machine, while providing the extra power associated with a pushdown stack memory. We provide an example of a useful control specification that cannot be implemented using a finite state machine, but can be implemented using an Exception Handling Controller.

A novel recurrent neurofuzzy network is developed in this paper. The network model is composed by two strucutres: a fuzzy system and a neural network. The fuzzy system contains fuzzy neurons modeled using t-norms and s-norms. The neural network is composed by nonlinear elements placed in series with the fuzzy system. The network model implicitly encodes a fuzzy rule-based system and its recurrent multilayered structure performs fuzzy inference. The topology induces a clear relationship between the network structure and the associated fuzzy rule-based system. Network learning involves three main steps. The first step uses a modified vector quantization approach to granulate the input universes. The next step assembles the network connections and their initial, randomly chosen weights. The third step uses two main paradigms to update the network weights: gradient descent and gradient projection method. The recurrent fuzzy neural network is particularly suitable to model nonlinear dynamic systems and to learn sequences. Computational experiment with a classic prediction problem benchmark shows that the fuzzy neural model outperforms a finite impulse response neural network.

In an ordered decision information system, dominance-based rough set approach (DRSA) were used to compute reducts of the system which preserve the lower and upper approximations of upward union and downward union of decision classes. Recently, class-based reducts (L-reduct and U-reduct) which preserve respectively lower and upper approximates of each decision classes was proposed. In this paper, we modify the definition of U-reduct and propose the concept of U*-reduct which is comprehensive and convenient for drawing decision rules. We give an equivalent description of L-reduct and U*-reduct, and compute the L-reduct and U*-reduct utilizing discernibility function defined by discernibility sets of objects.

In this paper, we deal with minimum spanning tree problems where each edge weight is a fuzzy random variable. In order to consider the imprecise nature of the decision maker's judgment, a fuzzy goal for the objective function is introduced. The problem is to seek a spanning tree maximizing the expectation of a degree of possibility or necessity that the fuzzy goal is satisfied. After showing that the problem to be solved is a minimum ratio spanning tree problem, a polynomial-time algorithm is provided to solve the problem.

In this paper, we consider the mixture of sparse linear regressions model. Let ${\beta}^{(1)},\ldots,{\beta}^{(L)}\in\mathbb{C}^n$ be $ L $ unknown sparse parameter vectors with a total of $ K $ non-zero coefficients. Noisy linear measurements are obtained in the form $y_i={x}_i^H {\beta}^{(\ell_i)} + w_i$, each of which is generated randomly from one of the sparse vectors with the label $ \ell_i $ unknown. The goal is to estimate the parameter vectors efficiently with low sample and computational costs. This problem presents significant challenges as one needs to simultaneously solve the demixing problem of recovering the labels $ \ell_i $ as well as the estimation problem of recovering the sparse vectors $ {\beta}^{(\ell)} $. Our solution to the problem leverages the connection between modern coding theory and statistical inference. We introduce a new algorithm, Mixed-Coloring, which samples the mixture strategically using query vectors $ {x}_i $ constructed based on ideas from sparse graph codes. Our novel code design allows for both efficient demixing and parameter estimation. In the noiseless setting, for a constant number of sparse parameter vectors, our algorithm achieves the order-optimal sample and time complexities of $\Theta(K)$. In the presence of Gaussian noise, for the problem with two parameter vectors (i.e., $L=2$), we show that the Robust Mixed-Coloring algorithm achieves near-optimal $\Theta(K polylog(n))$ sample and time complexities. When $K=O(n^{\alpha})$ for some constant $\alpha\in(0,1)$ (i.e., $K$ is sublinear in $n$), we can achieve sample and time complexities both sublinear in the ambient dimension. In one of our experiments, to recover a mixture of two regressions with dimension $n=500$ and sparsity $K=50$, our algorithm is more than $300$ times faster than EM algorithm, with about one third of its sample cost., Comment: To appear in IEEE Transactions on Information Theory

This paper investigates the effects of antenna correlations on spatial diversity and multiuser diversity. Using an upper bound on achievable capacity from order statistic theory, we quantify the interactions between spatial correlation, spatial diversity and multiuser diversity. Our theoretical analysis and simulation results demonstrate a positive influence of spatial correlations on achievable capacity of spatial diversity techniques combined with multiuser diversity, while it has been known that spatial diversity limits a multiuser diversity gain without spatial correlation.

This paper experimentally investigates the vibration faults of rotor, such as the unbalance, the loosening and the friction, using the rotor test rig. According to the theory of fractal and chaos, the vibration signal series are reconstructed. By the calculation of correlation dimension and Kolmogorov entropy, the dynamic characteristics of the system are then reproduced. The results show that the correlation dimension and Kolmogorov entropy of friction fault are the largest, that of the loosening fault are little lower, and that of the unbalance fault are the smallest. Different faults have different range of correlation dimension and Kolmogorov entropy, and they can be used to identify the fault symptoms.

This paper presents a case study addressing the comparison between different SAR polarimetric mode for tropical forest stratification: Full polarimetry (FP), Dual Polarimetry (DP) and Compact Polarimetry (CP). These 2 latter modes are simulated using FP data acquired by the L band PALSAR sensor over 2 study sites. Cayenne in French Guyana and the Fazenda Sao Nicolau in Brazil. The classification approach used to evaluate each mode based on Support Vector Machine (SVM) algorithm shows the good capabilities of the FP mode to discriminate different kinds of vegetation. The choice of one DP mode that gives twice bigger swath than FP mode depends on the study classes. In this tropical environment the hh/hv existing mode seems to be a good choice and CP mode show a good alternative to all the actual DP modes.

Various model-based control schemes, e.g. feedback-linearizing control of robotic manipulators with series elastic actuators require time derivatives of the control forces, i.e. of the equations of motion. This is referred to as the higher-order inverse dynamics problem. As this must be evaluated on the controller hardware, computationally efficient formulations are crucial. While recursive O (n) inverse dynamics algorithms are well-established, such for the higherorder inverse dynamics are relatively new. O (n) algorithms for the latter problem were published recently building upon the known recursive inverse dynamics algorithms in terms of Denavit-Hartenber (DH) parameters. The DH convention is restrictive, however. Also the higher-order formulations are naturally very complex. Both issues are addresses in this paper with concepts from screw and Lie group theory. An O (n) algorithm for determining the first and second time derivative of the inverse dynamics solution is presented. The Lie group approach provides a high level of compactness while ensuring the computational efficiency. Being frame invariant this formulation allows for using representations in different frame. Two formulations are presented, one using the body-fixed and the other the so-called hybrid representation of twists. The latter is deemed to be computationally more efficient that the classical body-fixed version. Results are shown for a 6-DOF industrial manipulator.

A novel vector quantizer (VQ) design algorithm for noisy channels is presented in this paper. The algorithm, termed fuzzy channel-optimized vector quantizer (FCOVQ) design algorithm, performs the codeword training using an optimal fuzzy clustering technique where the channel noise is taken into account. In the existing crisp channel-optimized VQ (CCOVQ) design algorithms, the quantization accuracy is traded for less sensitivity to channel noise. However, because of utilizing the optimal fuzzy clustering process for VQ design, the FCOVQ algorithm can effectively reduce the sensitivity to channel noise while maintaining the quantization accuracy. Therefore, given the same noisy channel, the FCOVQ can have better rate-distortion performance than that of the CCOVQ techniques.

In this paper, the main goal is to design an approach that performs fault detection and isolation in non linear systems. Fault diagnosis is established by regarding system as an interpolation of multiple LTI uncertainty models and not as a single global model. In multiple model framework, the purpose of the paper is the generation of a robust activation function of the representative linear model. The fault diagnosis method presented here is based on a bank of robust and decoupled parity spaces. The proposed method allows detection, isolation and estimation of multiple faults which appear simultaneously or in a sequential way. Robust model selection is obtained by considering the residual vector insensitive to faults, but sensitive to modelling uncertainty. Performances of the method are tested on a simulated example

Texture synthesis is a modern technology which grows rapidly in recent ten years. An improved example based texture synthesis algorithm is proposed. The basic idea is from the famous texture synthesis algorithm of WL00[4].Two improvements are done in this paper. The first one is to automatically determine the neighborhood size during the neighborhood match process, the second one is that each time three pixels other than one are synthesized to accelerate the synthesis procedure. Experimental results show that the proposed algorithm is effective and efficient.

This paper studies non-asymptotic model selection and recovery of sparse signals in high-dimensional, linear inference problems. In contrast to the existing literature, the focus here is on the general case of arbitrary design matrices and arbitrary nonzero entries of the signal. In this regard, it utilizes two easily computable measures of coherence—termed as the worst-case coherence and the average coherence—among the columns of a design matrix to analyze a simple, model-order agnostic one-step thresholding (OST) algorithm. In particular, the paper establishes that if the design matrix has reasonably small worst-case and average coherence then OST performs near-optimal model selection when either (i) the energy of any nonzero entry of the signal is close to the average signal energy per nonzero entry or (ii) the signal-to-noise ratio (SNR) in the measurement system is not too high. Further, the paper shows that if the design matrix in addition has sufficiently small spectral norm then OST also exactly recovers most sparse signals whose nonzero entries have approximately the same magnitude even if the number of nonzero entries scales almost linearly with the number of rows of the design matrix. Finally, the paper also presents various classes of random and deterministic design matrices that can be used together with OST to successfully carry out near-optimal model selection and recovery of sparse signals under certain SNR regimes or for certain classes of signals.

The problem of reconstructing digital images from degraded measurement is regarded as a problem of importance in various fields of engineering and imaging sciences. The main goal of denoising is to restore a noisy image to produce visually high quality image. In general, image denoising imposes a compromise between noise reduction and preserving significant image details. In this paper we propose to use multi-spinning based wavelet denoising for restoring images corrupted by Gaussian noise. Multi-spinning uses multiple cyclic shifted versions of image, where each version would capture more detail during decomposition. The visual and quantitative analysis suggest that the proposed method yields better results.

Novel single integral representations for the multivariate probability density function (PDF) and cumulative distribution function (CDF) of Gaussian class Weibull distributions are derived. A special linear transformation of independent Gaussian random variables is used to generate correlated Weibull random variables. The solutions are expressed in terms of familiar functions which are available in common mathematical software. The well known equal (constant) correlation model is used in this paper. The advantage of the new representation is that only a single integral computation is needed to compute a L-dimensional distribution. Some applications of the new distribution representations are also given.

Reliability is an important index to evaluate product quality, and reliability degree is most commonly investigated among various reliability charac¬teristic values. When generalized stress and generalized strength are fuzzy variables, this paper proposes a method for solution based on two algo¬rithms, i.e. the computation of fuzzy reliability using fuzzy number subtraction and fuzzy failure degrees. The advantages of the method lie in its uncomplicated and clear principle as well as low computational complex¬ity. Since one real number can be considered as a trigo¬nometric fuzzy number with its left and right distri¬buted parameters equivalent to zero, the method is applicable to the case of combination of fuzzy va¬riables and constant variables.