Xiaohu Ru

Single-Channel Blind Source Separation of Co-Frequency Overlapped GMSK Signals Under Constant-Modulus Constraints

As for the single-channel overlapped signals, in which the modulation parameters of the component signals are identical or similar, it is still a great challenge to extract the component signals. Aiming at this problem, a single-channel blind source separation (SCBSS) algorithm, constrained by constant modulus, is proposed for the overlapped Gaussian minimum shift keying (GMSK) signals and this algorithm also applies to other overlapped signals with constant modulus (CM), such as binary phase shift keying (BPSK) signals and quadrature phase shift keying(QPSK) signals. By tracking the channels of the component signals, the algorithm can effectively estimate their symbol sequences. Simulation results show that the proposed algorithm can work more robustly and reduce the frame error rate (FER), comparing with the competing algorithms.

Normalized residual-based outlier detection

Outlier detection is an important issue in data mining and knowledge discovery. The aim is to find the patterns that deviate too much from others. In this paper, a universal outlier detection method based on normalized residual is proposed. Different from previous methods, the residual of a pattern is calculated corresponding to its nearest normal patterns, so that the interaction between outliers is eliminated. To implement this, the method first estimates the center of normal patterns and derives the initial set of them, and then iteratively calculates the residual of the nearest pattern outside the set. Those with small residuals will be added to the set of normal patterns, and others are picked out as outliers. An effective distance weighting is also introduced to the calculation of the normalized residual. Simulation results show that the proposed method is efficient in detecting outliers and can hold a high detection probability even when 30% outliers appear in the dataset.

Emitter identification based on the structure of unintentional modulation

Unintentional modulation (UIM), which is unavoidable and unique to individual emitters, can be used to reliably realize emitter identification. Previous identification methods extract features from either some parts of the signal, ignoring the UIM on the other parts, or immediately the whole signal, resulting in heavy computational loads. In this paper, we take the structure of UIM into consideration, and propose a new feature extraction scheme. We first analyze the mechanism of UIM, realizing that the jitter frequency or intensity of UIM is time-varying throughout the signal. The parts of the signal with slow jitters are then located, in which the bandwidths of UIM are actually much smaller than the initial sampling rate; thus, down-sampling is applied to these parts. Lastly, features extracted from different parts are combined. Experiments on real-world data validate the superior recognition performance and computing speed of the proposed feature extraction scheme.

Near-optimal estimation of radar pulse modulation waveform

In this study, the authors focus on pulse modulation waveform estimation based on a sequence of intercepted pulses from the same radar emitter. In general, modulation waveform estimators first perform pulse alignment in time and frequency separately, and then accumulate the aligned pulses to estimate the waveform. However, commonly used alignment methods may lead to considerable alignment errors which are difficult to detect and compensate, especially under low signal-to-noise ratio (SNR) conditions, resulting in non-ideal effects of accumulation. This study proposes a robust and nearly optimal modulation waveform estimation algorithm. The new algorithm first aligns pulses in time and frequency jointly via the cross-ambiguity function to avoid the transfer and accumulation of alignment errors. After that, an iterative maximum-likelihood estimator is invoked to achieve the waveform estimation. Theoretical analysis and extensive experiments show that the proposed algorithm has much smaller alignment errors and better modulation waveform and modulation parameter estimation ability than competing methods at low SNRs, and can approach the ideal case. Moreover, this algorithm does not make any assumption on the type of modulation and is computationally efficient, thus having broad applications.

Class discovery based on K-means clustering and perturbation analysis

Class discovery, which aims to identify the underlying category structure, is an important issue in pattern recognition and knowledge discovery. The key task in class discovery is to estimate the number of classes. Classical estimation approaches usually face the problems of low accuracy, high complexity, or difficulty in choosing an appropriate penalty function. In this paper, an effective class discovery method is proposed. The method first utilizes the characteristics of the mean-square-error produced by k-means clustering, giving a coarse estimate of the number of classes, and then calculates the difference between the clustering results obtained from the original dataset and the perturbed dataset to further determine the real number of classes. Experiments on simulated and real-world data demonstrate that the proposed method has satisfactory performance in different situations. Moreover, this method relies loosely on artificially selected parameters, thus can be reliably used in wide applications.