In this study, the authors focus on the state estimation of a non-linear system described by a Takagi–Sugeno multiple model submitted to unknown inputs and outputs. The proposed approach consists of a mathematical transformation which enables to consider the unknown outputs as unknown inputs that can be eliminated by a designed multiple observer. To evaluate the efficiency of the proposed approach, the convergence conditions of the state estimation error are formulated as linear matrix inequalities. Simulation examples are given to illustrate the proposed methods.

A two-stage recursive least-squares identification method and a two-stage multi-innovation stochastic gradient method are derived for Box–Jenkins (BJ) systems. The key is to decompose a BJ system into two subsystems, one containing the parameters of the system model and the other containing the parameters of the noise model, and then to estimate the parameters of the system model and the noise model, respectively. The simulation examples indicate that the proposed algorithms can generate highly accurate parameter estimates and require small computational burden.

Spectrum sensing is a key component to realise the cognitive radio. The main requirements of spectrum sensing are the prediction of signal status in multiple frequency bands in a low signal-to-noise ratio (SNR) and decision reliability. This study proposes a novel multinode wideband sensing technique to predict the status of multiple frequency bands based on the integration of entropy and cyclic properties of received signals. It uses the uncertainty and auto-correlation properties of the deterministic signal and noise in the frequency domain for signal detection. To increase the decision reliability, cooperative sensing techniques are being used for spectrum sensing. Although cooperation among multiple cognitive users enhances the sensing performance, presence of few suspicious/malicious cognitive users severely degrade the decision reliability of the system. Hence, in this work, generalised extreme studentised deviate and adjusted box-plot methods are introduced to eliminate multiple malicious users in the cooperation. The proposed sensing method shows the best performance and is less severe to noise uncertainties compared to the traditional sensing methods in the literature. It enhances the sensing performance by 2.5 dB using five nodes in cooperation for same sensing parameters compared to other detection methods. It is a significant improvement for IEEE 802.22 systems that work under low SNR environment.

A target localisation estimator based on time difference of arrival (TDOA) measurement is proposed. The localisation estimator is designed in the framework of the recently developed robust least-squares (RoLS) estimator, which provides an unbiased estimation result and can be implemented with a recursive filtering structure. However, when the RoLS estimator is applied to the localisation problem, its localisation performance depends on the knowledge of the stochastic information of the TDOA measurement. This dependency means that incorrectly given information causes localisation error. Therefore to complement the dependency of the given stochastic information of the TDOA measurement, we design a compensation procedure based on the constraints on the state variables of the estimator. The performance of the proposition under several cases of incorrectly given stochastic information is verified through computer simulation, and its filtering structure is compared with other existing localisation algorithms mathematically. In addition, the entire process of the proposed localisation estimator is derived as a recursive form for real-time applications.

In this study, the authors investigated some new inequalities on sparse representation for pairs of bases and frames, which would enrich the theory ensemble. First, for fixed pairs of bases, frames and the signal to be represented, we presented the bounds (which can be used in practice directly) of l ₀-norm of signal coefficients by the max/min cross-inner-products, as is of much importance to bases and frames selections in sparse representation. Also, the error bounds associated with the given min cross-inner-products are achieved. Moreover, the equivalence condition between l ₀ solution and l ₁ solution was achieved via relations of cross-inner-products, as can be applied directly for selection of optimal bases. Finally, the estimation of the parameters of cross-inner-products was shown.

In this study, a novel technique that recovers the temporal structure of speech power spectrum is proposed. The histogram of average speech log power spectrum shows that the contamination of noise leads to the shift of noise peak, which in return degrades the performance of speech recognition systems. A two-step scheme is proposed to weaken the noise effects by first reducing the noise variance and then shifting the noise mean. The proposed algorithm consists of two parts, two-dimensional smoothing and controlled noise subtraction, which leads to the name SNS. The proposed algorithm manages to solve the speech probability distribution function discontinuity problem caused by traditional spectral subtraction series algorithms. In contrast to the clean speech estimation methods, the proposed algorithm does not need a prior speech/noise statistical model, which makes it simple but effective. The effectiveness of the proposed filter is tested using the AURORA2 database. Very promising results are obtained, 88.59% for noisy speech (average from signal-to-noise ratio 0–20 dB). Comparison is made against eight state-of-the-art speech recognition algorithms. Overall the proposed algorithm produces significant improvements over the comparison targets.

Having many benefits, phase-mode transformation technique is used frequently in beamforming with circular arrays. Owig to the fact that this transformation is frequency dependent, some techniques are used to make this transformation appropriate for broadband applications. When this technique is used with uniform circular arrays, its performance degrades because of the approximations applied to the formulations. In this study, the authors propose a new beamforming method not only robust against uncertainty in the steering vector of the desired signal, but also robust against errors caused by the phase-mode transformation. The authors assume that steering vector of the desired signal pertains to a ball set and ellipsoid sets are considered to cover the transformation matrix columns. The proposed beamforming method is based on semi-definite programming. The performance of the proposed beamforming method is compared with several other beamforming techniques by using a simulation study.

Recently, speech scientists have been motivated by the great, success of building margin-based classifiers, and have thus proposed novel methods to estimate continuous-density hidden Markov model (HMM) for automatic speech recognition (ASR) according to the notion that the decision boundaries determined by the estimated HMMs attain the maximum classification margin as in learning support vector machines. Although a good performance has been observed, the margin used in the ASR community is often specified as a parameter that has no explicit relationship with the HMM parameters. The issues of how the margin is related to the HMM parameters and how it directly characterises the generalisation capability of HMM-based classifiers have not been addressed so far in the community. In this study, the authors attempt to formulate the margin used in the soft margin estimation framework as a function of the HMM parameters. The key idea is to relate the standard distance-based margin with the concept of divergence among competing HMM state Gaussian mixture model densities. Experimental results show that the proposed model-based margin function is a good indication about the quality of HMMs on a given ASR task without the conventional needs of running experiments extensively using a separate set of test samples.

In this study, the authors present the research results on the robust investigate the robust l ₂ –l_∞ filtering for Markovian jump linear systems with multiple sensor faults, uncertain probability transition matrix and time-varying delays. The multiple sensor faults are modelled as multiple independent Bernoulli processes with constant probabilities. The uncertain probability transition matrix is modelled via the polytopic uncertainties for each row in the transition matrix. By using the augmentation method, the filtering error system with stochastic variables is derived. Since of the stochastic variables, the traditional stability condition is not qualified for the analysis of the filtering error systems. Thus, the exponentially mean-square stability and the robust l ₂ –l_∞ performance are adopted for the filtering error system. By choosing the Lyapunov-based method, sufficient conditions which can guarantee the exponentially mean-square stability and the robust l ₂ –l_∞ performance are obtained in the forms of matrix inequalities. Based on these conditions, the filter design method is proposed and the estimator parameters can be obtained by solving a set of linear matrix inequalities. Finally, a numerical example with two modes is used to show the design procedure and the effectiveness of the proposed design approach.

In this study, an improved method based on evolutionary algorithms for denoising of satellite images is proposed. In this approach, the stochastic global optimisation techniques such as Cuckoo Search (CS) algorithm, artificial bee colony (ABC), and particle swarm optimisation (PSO) technique and their different variants are exploited for learning the parameters of adaptive thresholding function required for optimum performance. It was found that the CS algorithm and ABC algorithm-based denoising approach give better performance in terms of edge preservation index or edge keeping index (EPI or EKI) peak signal-to-noise ratio (PSNR) and signal-to-noise ratio (SNR) as compared to PSO-based denoising approach. The proposed technique has been tested on satellite images. The quantitative (EPI, PSNR and SNR) and visual (denoised images) results show superiority of the proposed technique over conventional and state-of-the-art image denoising techniques.

In order to meet the bandwidth need of modern wideband digital reconnaissance receiver, an implementation based on compressed sensing is proposed. The compressed sensing method is directly used to sample and reconstruct multi-band RF signal in this receiver. The original signal is mixed with Bernoulli random signal, and then filtered by low-pass filter. As completing the multiple narrow-band signals recovery within broadband range, the sampling signal is reconstructed in the digital domain. In this study, a novel reconstruction algorithm is proposed, which is adaptive conjugate gradient pursuit multiple measurement vectors (ACGPMMV), to overcome the drawback of orthogonal matching pursuit multiple measurement vectors (OMPMMV). Meanwhile, how to reduce the channel number of system required is also analysed, which will be reduce hardware costs. The shortcoming of limitation of sampling chip's analogue bandwidth in the parallel alternating sampling system is overcame, and the requirement of receiver bandwidth is achieved by using the lower rate sampling. In this study, the study is carried out by means of numerical simulations of multi-signals in the different system model. The conclusions illuminate the algorithm can get well-signal reconstruction results under the conditions of less channel and low sampling rate, and well-noise stability.

In this study, a low-power and high performance architecture for finite impulse response digital filter based on the ring topology which is modelled from recurrent neural network is presented. The proposed structure is based on a ring topology reduced number of multipliers, adders and also CLK cycles. In the design, all the operators including multipliers and adders have been designed at gate level. Multiplication is a very important operation in many digital filters hence, the authors designed a novel and modified retiming serial multiplier. To increase the performance, the authors use two types of adders, a proposed high-speed logarithmic carry look ahead adder and a carry save adder with four inputs. The proposed structure is modelled and verified using FPGA and simulation results. It has been successfully synthesised and implemented with Xilinx ISE 7.1 and Virtex IV FPGA, target device Xc4vf100. The results demonstrate that the proposed method has high performance and low-power consumption.

This study provides an event-triggered ℋ _∞ filtering design method for a discrete linear system under network environments. First, a periodic event-triggered communication scheme and a networked filter error system model for networked ℋ _∞ filtering are presented. In this scheme and model: (i) the sensor is time-triggered; (ii) the transmitter is event-triggered in a periodic manner; and (iii) the closed-loop system is modelled as a time-delay dependent filter error system. Second, under consideration of the proposed communication scheme, an ℋ _∞ filtering analysis criterion and a stabilisation criterion are derived. Compared with those where the communication scheme and the filter must be individually designed in some existing ones, the communication and filtering parameters can be obtained simultaneously. In particular, a co-design algorithm is provided to obtain the communication and filtering parameters in a unified framework for using less network bandwidth. Finally, two examples are given to show the advantages of the proposed method.

The dynamic behaviour of many systems can be approximated by a static non-linearity in series with a linear dynamic part. Systems with static input or output non-linearities are very common in many engineering applications. Such models are known as block-oriented models in the existing literature. The Hammerstein model is a special kind of block-oriented model, where a non-linear block is followed by a linear system. This study investigates the identification of Hammerstein systems with asymmetric two-segment piecewise-linear non-linearities. The basic idea is to employ a key-term separation technique and a corresponding auxiliary model initially. Then, the identification problem of non-linear system is changed into a non-linear function optimisation problem over parameter space. Further, the estimates of all the parameters of the non-linear block, the linear subsystem and the noise part are obtained based on an improved particle swarm optimisation algorithm. Finally, simulation examples are included to demonstrate the effectiveness and robustness of the proposed identification scheme.

Sparsity is one of the key points in the compressed sensing (CS) theory, which provides a sub-Nyquist sampling paradigm. Nevertheless, apart from sparsity, structures on the sparse patterns such as block structures and tree structures can also be exploited to improve the reconstruction performance and further reduce the sampling rate in CS framework. Based on the fact that the block structure is also sparse for a widely studied block sparse signal, in this study, a double-level binary tree (DBT) hierarchical Bayesian model is proposed under the Bayesian CS (BCS) framework. The authors exploit a recovery algorithm with the proposed DBT structured model, and the block clustering in the proposed algorithm can be achieved fastly and correctly using the Markov Chain Monte Carlo method. The experimental results demonstrate that, compared with most existing CS algorithms for block sparse signals, our proposed DBT-based BCS algorithm can obtain good recovery results with less time consuming.

Prognostic utility of microvolt T-wave alternans (TWAs) has been established since its clinical acceptance as markers for malignant ventricular arrhythmias, leading to sudden cardiac death. Accurate detection of TWA from surface electrocardiography is a challenge because of invisible nature of the phenomenon. A novel TWA detection scheme based upon analysis of continuous-time wavelet entropy (CTWE) trend of consecutive ventricular repolarisation complexes is presented. The CTWE is computed using relative wavelet energy coefficients of continuous wavelet transform. Variety of simulated alternan waveforms, wavelet functions, frequency bands and noise levels are used to test the algorithm. The algorithm achieves a sensitivity of 100% at signal-to-noise ratio (SNR) >35 dB for all the selected wavelet functions and sensitivities of 99.5, 97 and 92% for Symlet4, Mexican Hat and truncated Morlet functions, respectively, at 30 dB SNR. A performance improvement of 5 dB is achieved by only computing the wavelet coefficients at the optimal frequency band. This study concludes that CTWE can successfully characterise the heterogeneity of cardiac repolarisation and detect TWA phenomenon.