Limin Yu

Optimum Receiver Design for Broadband Doppler Compensation in Multipath/Doppler Channels With Rational Orthogonal Wavelet Signaling

In this paper, we address the issue of signal transmission and Doppler compensation in multipath/Doppler channels. Based on a wavelet-based broadband Doppler compensation structure, this paper presents the design and performance characterization of optimum receivers for this class of communication systems. The wavelet-based Doppler compensation structure takes account of the coexistence of multiple Doppler scales in a multipath/Doppler channel and captures the information carried by multiple scaled replicas of the transmitted signal rather than an estimation of an average Doppler as in conventional Doppler compensation schemes. The transmitted signal is recovered by the perfect reconstruction (PR) wavelet analysis filter bank (FB). We demonstrate that with rational orthogonal wavelet signaling, the proposed communication structure corresponds to a Lth-order diversity system, where L is the number of dominant transmission paths. Two receiver designs for pulse amplitude modulation (PAM) signal transmission are presented. Both receiver designs are optimal under the maximum-likelihood (ML) criterion for diversity combination and symbol detection. Good performance is achieved for both receivers in combating the Doppler effect and intersymbol interference (ISI) caused by multipath while mitigating the channel noise. In particular, the second receiver design overcomes symbol timing sensitivities present in the first design at reasonable cost to performance.

A New Contour Detection Approach in Mammogram Using Rational Wavelet Filtering and MRF Smoothing

This paper presents a new approach to detect breast contour in mammogram. Formed from a class of rational orthogonal wavelets (ROWs), a 2-D image filter is constructed for prefiltering of the mammogram. The filtered image facilitates a simple binarisation of the mammogram. An initial breast contour is then extracted from the binarised image with a simple boundary scan technique. Based on Markov Random Field (MRF) modelling and iterated conditional modes (ICM) relaxation, a smoothing algorithm is developed to further smooth the initial contour. The proposed smoothing algorithm has a unique advantage of smoothing the breast contour while the nipple is preserved with high fidelity. In comparison with contour detection techniques relying on the calculation of varying thresholds based on histgram analysis, a single fixed ROW image filter is sufficient for all mammograms being analysed. The ROW filter is adaptive to varying statistics of mammograms regarding the pixel intensity. Results prove the robustness of the proposed detection algorithm for 82 mammograms from the Mini-MIAS database.

Broadband Doppler compensation for rational wavelet-based UWA communication systems

This paper proposes a broadband Doppler compensation structure based on rational orthogonal wavelet signaling and perfect reconstruction wavelet filter banks. A transmultiplexer system model is presented to characterise wavelet signalling, channel transmission and Doppler compensation processes. Because the underwater acoustic (UWA) Doppler scales are rational numbers very close to 1, a special class of wavelets, the rational orthogonal wavelets with a scale dilation factor of aj : j isin Z, 1 < a < 2, are introduced in this application. These rational wavelets are well adapted with the proposed Doppler compensation structure and achieve good system performance. By effectively resolving the multiscale nature of the UWA channels for wideband signal propagation, this class of wavelets forms optimal shaping pulses with unique robustness against Doppler dispersion and has a potential for further development towards a new class of rational wavelet based UWA systems

Broadband passive sonar detection using rational orthogonal wavelet filter banks

A broadband passive sonar detector based on rational orthogonal wavelet filter banks (FBs) is proposed. This wavelet-based subband energy detection approach is able to detect an emission with unknown frequency content in severe multipath and noisy ocean environment. A geometrical acoustic channel model is adopted to synthesise the echo from a moving source using ray tracing. The performance of the wavelet FB detector is analysed under designated detection scenarios with a single detector and three different random emissions varying in bandwidth, source motion velocity and direction. Two different multipath scenarios, severe multipath and one-dominant-path scenario, are simulated by modifying the depth of the detector. The receiver operating characteristic (ROC) curves are generated for the wavelet FB detector and are compared with the ROC curves of a conventional energy detector (CED) under the same scenario.

Sensor network traffic load prediction with Markov random field theory

Following recent advances in wireless communications and computing technology, sensor networks are widely deployed in different fields for both monitoring and control purposes. In this work, we focus on using Markov random field (MRF) theory to model traffic intensity of the three types of sensor networks. Shortest path routing is adopted in the three typical lattice network models. Then, the influences, which affect the traffic distribution dynamically in real situations, are modelled by adding the Gaussian noise to the traffic load distribution in the MATLAB simulation. Given measurements of real-time samples of traffic, we are able to predict the traffic at each sensor node for specific network models by a MRF smoothing algorithm.

Mammogram Mass Classification with Temporal Features and Multiple Kernel Learning

Based on previous work on regional temporal mammogram registration, this study investigates the combination of image features measured from single regions (single features) and image features measured from the matched regions of temporal mammograms (temporal features) for the classification of malignant masses. Three SVM kernels, the multilayer perceptron kernel, the polynomial kernel, and the gaussian radial basis function kernel, and the combination of these kernels, the multiple kernel learning method, were applied to both single and temporal features for the mass classification. To combine the two types of features, 3 combination rules, Linear combination, Max and Min, were used to combine classification results obtained on single and temporal features. The results showed that combining the MKL classification results on single features, and MKL classification results on temporal features, with Min rule produces the best classification results. The experiment result indicates that incorporating the temporal change information in mammography mass classification can improve the performance detection.

Temporal change analysis for computer aided mass detection in mammography

This paper presents a method to extract change information from temporal mammogram pairs and to incorporate the temporal change information in the malignant mass classification. In this method, a temporal mammogram registration framework which is based on spatial relations between regions of interest and graph matching was used to create correspondences between regions of current mammogram and regions of previous mammogram. 18 image features were then used to capture the differences (temporal changes) between the matched regions. To assess the contribution of temporl change information to the mass detection, 4 methods were designed to combine mass classification on image features measured on single regions and mass classification on temporal features to improve overall mass classification. The method was tested on 95 pairs of temporal mammograms using k-fold cross validation procedure. The experimental results showed that, when combining two classification results using linear combination or by taking minimum value, the Az score of overall classification performance increased from 0.8843 to 0.8958 and 0.8962 respectively. The results demonstrated that registering temporal mammograms, measuring temporal changes from matched regions and incorporating the change information in the mass classification improves the overall mass detection.

Thresholding strategy in passive detection via wavelet filter banks

In this paper, we evaluate thresholding algorithms applied in wavelet filter bank (FB) based passive detection in radar and sonar systems. Addressing a random emission with unknown parameters, the emission and the received signal are modelled by a Gaussian process with unknown mean and variance. The wavelet FB decomposes the received signal into subchannels. A decision on the absence and presence of the emission is then made by thresholding against the FB outputs. Formulae are derived to prove the detection performance using different thresholding strategies. A graphical comparison of the detection performances with different strategies and in different scenarios is presented. Design principles are derived under this wavelet FB based detection framework.