Shilian Wang

Aspect-Aware Target Detection and Localization by Wireless Sensor Networks

This paper considers the active detection of a stealth target with aspect dependent reflection (e.g., submarine, aircraft, etc.) using wireless sensor networks (WSNs). When the target is detected, its localization is also of interest. Due to stringent bandwidth and energy constraints, sensor observations are quantized into few-bit data individually and then transmitted to a fusion center (FC), where a generalized likelihood ratio test (GLRT) detector is employed to achieve target detection and maximum likelihood estimation of the target location simultaneously. In this context, we first develop a GLRT detector using one-bit quantized data which is shown to outperform the typical counting rule and the detection scheme based on the scan statistic. We further propose a GLRT detector based on adaptive multi-bit quantization, where the sensor observations are more precisely quantized, and the quantized data can be efficiently transmitted to the FC. The Cramer-Rao lower bound (CRLB) of the estimate of target location is also derived for the GLRT detector. The simulation results show that the proposed GLRT detector with adaptive 2-bit quantization achieves much better performance than the GLRT based on one-bit quantization, at the cost of only a minor increase in communication overhead.

Sequential distributed detection via level-triggered sampling in underwater wireless sensor networks

In implementations of distributed detection based on underwater wireless sensor networks (UWSN), a few technique challenges are faced. Among them is the strict limited communication ability of local sensors due to constrained bandwidth as well as energy, and the severe synchronization error among sensors which makes data fusion in the fusion center (FC) difficult. To solve the above problems to some extent, a sequential distributed detection (SDD) scheme based on level-triggered sampling (LTS) mechanism is proposed. In the LTS-SDD scheme, sequential test method is applied in both sensors and the FC, then the distributed detection system works with asynchronous sampling among sensors and low-speed communication between sensors and the FC. Though potential, the proposed scheme faced some unaddressed challenges when used in practical underwater distributed detection system, which are also discussed in the paper. To validate the suggested scheme, distributed detection of weak single frequency signal via UWSN is studied under some assumptions. Corresponding simulation results show that the LTS-SDD scheme works well and its error performance as well as average decision delay performance is superior to sequential detection (SD) scheme with single sensor.

Improved particle swarm optimization algorithm of clustering in underwater acoustic sensor networks

Aiming at improving the poor stability and unsatisfactory clustering results of the existing underwater acoustic sensor networks (UASNs) clustering algorithms, this paper proposes a new clustering model. In the model the required transmission power of sensor nodes, as well as the cluster head residual energy and the cluster head loads are among consideration. With the clustering model, we design a novel clustering algorithm based on the discrete particle swarm optimization algorithm (PSO). We use the proposed clustering algorithm to cluster UASNs periodically with the cluster head being rotated dynamically. Simulation results demonstrate the reduction in network energy consumption and the prolonged network lifetime. Moreover it achieves higher stability.

Specific Emitter Identification based on nonlinear complexity of signal

Specific Emitter Identification (SEI) is to identify the emitters with various RF fingerprints, originated from the nonlinearity of the emitter power amplifiers. This paper firstly develops an improved Approximate Entropy (imApEn) algorithm, by modifying the tolerance interval of Approximate Entropy (ApEn), to extract the nonlinear complexity of the signals as a new steady-state RF fingerprint. Then a novel identification algorithm is proposed based on the combination of the EMD and the imApEn, which utilizes some independent RF fingerprints to form multi-dimensional feature space and then the emitter classification is performed by the support vector machine. Additionally, the noise immunity and robustness of the imApEn are evaluated via Logistic map with different parameters. Computer simulations are conducted under additive white Gaussian noise as well as impulsive noise and the results demonstrate that the proposed algorithm significantly out-performs the energy-entropy algorithm and correlation algorithm based on the Hilbert-Huang Transform.

Stochastic signal detection based on changeable fractional lower order moments in alpha stable noise

Non-Gaussian noises usually fail many conventional and effective signal detection techniques including the energy detector and the eigenvalue-based detector. The fractional lower order moment (FLOM) based detector has proved to be useful for unknown stochastic signal detection in α-stable distributed noises. However, the fixed exponent prevents the improvement of its performance. This paper presents a novel signal detection method based on changeable fractional lower order moments in non-Gaussian noise modeled by the α-stable distribution. The proposed detector would require the estimation of the characteristic exponent (α) and the dispersion (γ) of the background noises, to decide a proper bound using an empirical formula for piecewise processing. Computer simulations and field experiments are conducted to obtain the detection probabilities and ROC curves of the proposed detector, against the FLOM detector and Cauchy detector, in terms of the generalised signal-to-noise ratio and the characteristic exponent (α). Results show that the changeable fractional lower order moment detector significantly outperforms the FLOM based detector and Cauchy detector for small values of α and the simple implementation makes it an attractive solution for signal detection in α-stable distributed noises.

A novel Automatic Modulation Classification method based on Stockwell-transform and energy entropy for underwater acoustic signals

Automatic Modulation Classification (AMC) of communication signals plays a significant role in communication systems. However, conventional methods of modulation classification have poor performance in a shallow water environment. Recently, the Stockwell-transform (S-transform), a new time-frequency analysis method, receives widely attention in different areas. In this paper, we introduce the S-transform into modulation classification and propose a novel method of modulation classification under underwater acoustic channel. Firstly, we set up a model of underwater acoustic channel based on Bellhop and the multipath Rayleigh fading channel model. Next, we extract features of energy entropy of S-transform time-frequency spectrum of signals, and then input them into the classifier, Support Vector Machine (SVM). Meanwhile, different signal sets are considered, which have different number of signal schemes. Finally, Matlab simulating experiments are performed to evaluate the performance of the proposed method for each signal set under AWGN channel, and results show that the proposed method reaches higher probability of correct classification than convention methods. Aiming at the problem under multipath fading channel, especially underwater acoustic channel, the simulated results show it effectiveness.

Simulation study on RFI signal in impulse UWB radar system with equivalent time sampling

This paper focuses on the RFI signal in the impulse UWB system with equivalent time sampling. RFI signals in the equivalent time sampling cannot be completely recovered. To reveal the properties of the aliased RFI signals, we setup a simple mode of the RFI signal with time-varying phase and analyze the effect of the classical average operation. Simulations about the random and periodic equivalent time sampling schemes are performed and compared, and indicate that the RFI suppression has more relationship with the average factor other than the random equivalent time sampling scheme.