Optimal Detection and Performance of Distributed Sensor Systems

Global optimization of a distributed sensor detection system withfusion is considered, where the fusion rule and local detectors aresolved to obtain overall optimal performance. This yields coupledequations for the local detectors and the fusion center.The detection performance of the distributed system with fusionis developed. The globally optimal system performance is comparedwith two suboptimal systems. Receiver operating characteristics(ROCs) are computed numerically for the problem of detecting aknown signal embedded in non-Gaussian noise.     

Coherent detection in Laplace noise

The discrete-time detection of a time-varying, additive signal in independent Laplace noise is considered. Previous efforts in this area have been restricted to the constant signal, and identically distributed noise case. Theoretical (closed form) expressions for the false alarm and detection probabilities are developed for both the Neyman-Pearson optimal detector and the classical matched filter detector. Comparisons between the two detectors are made which illustrate the effects of signal-to-noise power ratio and sample size for certain false alarm and detection probability constraints. In view of the fact that the optimal Laplace detector is not UMP, we also investigate the effect of signal amplitude mismatch     

Structures for radar detection in compound Gaussian clutter

The problem of coherent radar target detection in a background of non-Gaussian clutter modeled by a compound Gaussian distribution is studied here. We show how the likelihood ratio may be recast into an estimator-correlator form that shows that an essential feature of the optimal detector is to compute an optimum estimate of the reciprocal of the unknown random local power level. We then proceed to show that the optimal detector may be recast into yet another form, namely a matched filter compared with a data-dependent threshold. With these reformulations of the optimal detector, the problem of obtaining suboptimal detectors may be systematically studied by either approximating the likelihood ratio directly, utilizing a suboptimal estimate in the estimator-correlator structure or utilizing a suboptimal function to model the data-dependent threshold in the matched filter interpretation. Each of these approaches is studied to obtain suboptimal detectors. The results indicate that for processing small numbers of pulses, a suboptimal detector that utilizes information about the nature of the non-Gaussian clutter can be implemented to obtain quasi-optimal performance. As the number of pulses to be processed increases, a suboptimal detector that does not require information about the specific nature of the non-Gaussian clutter may be implemented to obtain quasi-optimal performance     

Detection in Laplace Noise

The discrete time detection of a known constant signal in white stationary Laplace noise is considered. Exact expressions describing the performance of both the Neyman-Pearson optimal detector and the suboptimal linear detector are presented. Also, graphs of the receiver operating characteristics are given. The actual performance of the Neyman-Pearson optimal detector is compared to that predicted by a Gaussian approximation to the distribution of the test statistic.     

Performance of a Two-Sample Mann-Whitney Nonparametric Detector in a Radar Application

The detection of signals in an unknown, typically non-Gaussian noise environment, while attempting to maintain a constant false-alarm rate, is a common problem in radar and sonar. The raw receiver data is commonly processed initially by a bank of frequency filters. The further processing of the outputs from the filter bank by a two-sample Mann-Whitney detector is considered. When the noise statistics in all filters are identical, the Mann-Whitney detector is distribution free, i. e., the false-alarm probability may be prescribed in advance regardless of the precise form of the noise statistics. The primary purpose of this paper is to demonstrate the potential advantage of nonparametric detectors over conventional detectors. The signal detection performance of the Mann-Whitney detector is compared to that of an ordinary linear envelope detector plus integrator in the presence of Gaussian and several hypothetical forms of non-Gaussian noise. This comparison is made for both uniform and nonuniform distributions of noise power across the filter bank. Besides providing a much more constant false-alarm rate than the conventional detector, the Mann-Whitney detector's signal detection performance is found also to be much less sensitive to the form of the noise statistics. In one case, its detection sensitivity is found to be 11 dB better than that of the conventional detector. Even when the noise power density is made moderately nonuniform across the filter bank, the detection performance of the Mann-Whitney detector is found not to be significantly affected.     

Detectors for Multiomial Input

The binary detection problem is considered. Under an arbitrary noise environment, the input sample space can be transformed into a multinomial vector. Based on observations of this vector, the Neyman-Pearson optimal detector is developed for a known signal. When the signal strength is unknown, the likelihood ratio principle is followed to obtain consistent tests which use the Pearson's chisquare statistic. The resulting detectors are compared to others in terms of asymptotic relative efficiency under some actual noise distributions.     

拉普拉斯白噪声下的分组Turbo码

目前,传统航空测控通信所采用的纠错码大多是建立在高斯信道基础上的。然而,航空测控环境中不可避免存在着多种尖锐的噪声,测控通信纠错码的可靠性能在非高斯信道中尚未得到充分的研究。分析了一类国际航空遥测的分组Turbo码（BTC）在拉普拉斯白噪声信道下的译码和性能。将传统Chase迭代译码算法引入到拉普拉斯白噪声信道中,建立相应的数学模型,同时,基于该数学模型设计了3种不同的译码接收器下的BTC译码方案。仿真结果验证了该数学模型的正确性与可行性,在误码率为10^-4时最佳译码方案相比于硬限幅接收机有3.7 dB的增益,相比原有的高斯信道下的接收机仅有0.6 dB的性能损失。     

Modified Savage and Modified Rank Squared Nonparametric Detectors

A modified form of the basic Savage statistic is considered and the performance of a modified Savage (MS) nonparametric detector using this modified statistic is derived. Also, a detector using a modified rank squared statistic (MRS) is introduced. The asymptotic relative efficiency (ARE) of the detectors is determined for chisquare, Rician, and log-normal signal fluctuations when the background noise is assumed Gaussian. The ARE performance of the generalized sign (GS) and Mann-Whitney (MW) detectors is also determined for these families of fluctuations. The ARE performance of the various detectors is then compared, and the results of a computer simulation are presented in which, for a finite number of samples, the performance of the modified detectors is compared with the performance of the GS and MW detectors. It is shown that when using a large number of reference noise samples, the ARE of the GS and MW detectors, the MRS and RS detectors, and the MS and Savage detectors are 0.75, 0.868, and 1, respectively. It is also shown that when using a finite number of reference noise samples the MS and MRS detectors can give a superior performance to that obtained with the MW detector, and that this is particularly true in the cases in which the degree of signal fluctuation is high.     

Comments on `Detector relative efficiencies in the presence ofLaplace noise' by M.I. Dadi and J.R. Marks II

In the above-titled paper (see ibid., vol.AES-23, p.568-82, July 1987) M.I. Dadi and J.R. Marks II studied the relative efficiencies of the Neyman-Pearson optimal detector with respect to the linear and sign detectors, for the detection of a constant signal in additive Laplace noise. By applying the central limit theorem, they derived expressions for three types of asymptotic relative efficiencies (AREs). However, as noted in the above paper, the Gaussian approximation to the sign detector fails to yield the correct asymptotic efficiency. The commenter derives the correct ARE of the optimal detector with respect to the sign detector for the Laplace noise     

Detectability of signals in Laplace noise

Closed-form expressions for the false-alarm and detection probabilities attained by the optimum and the linear detectors of a positive signal in n independent samples of noise having a bilateral exponential or Laplace distribution require lengthy computation when n is large, and those for the optimum detector suffer from round-off error because their terms alternate in sign. It is shown how the method of saddlepoint integration can be conveniently applied to compute these probabilities, and numerical comparisons of the accuracies of the methods are presented. The relative efficiency of the two detectors is calculated as a function of n and found to approach the asymptotic value of 2 very slowly     

Robust detection of small stochastic signals

We consider the problem of detecting a stochastic signal in white not-necessarily-Gaussian noise, using vector valued observations. The locally optimal detector is presented and its performance evaluated. The least-favorable signal spectrum and noise density (over specified classes) are found, and it is shown that the detector using these least-favorable assumptions is minimax robust. The class of spectra is that of any stochastic signal of specified power whose spectrum can be bounded from above and from below by two given positive functions. The class of densities is the ε-contamination model. We present examples of the performance achievable with the robust detector in one of these the spectral uncertainty class corresponds to the unknown Doppler shift of a radar return signal. It is demonstrated that the standard matched-filter's performance degradation with increasing Doppler shift can be avoided almost entirely through use of the robust processor     

A noncoherent adaptive detection technique

An adaptive detection technique suitable for both stationary and nonstationary noise environments based upon a generalized likelihood ratio test (GLRT) formulation is presented. The detector, which is statistically equivalent to a special form of the Wilks's lambda test, noncoherently combines the information contained in a pulse train of arbitrary length for decision-making purposes. The probability density function of the test under the noise only hypothesis is shown to be central χ². Under the signal plus noise hypothesis, an exact statistical characterization of the test cannot be obtained, and, therefore, a Chernoff bound is derived. Results in terms of the probability of detection versus signal-to-noise ratio (SNR) obtained from Monte Carlo simulation, the Chernoff bound, and the optimal matched filter case are examined. The performance of the noncoherent detector is shown to be a function of the covariance matrix estimate and the number of data samples     

Optimum Spatial Processing of Passive Sonar Signals

The spatial structure of the likelihood ratio array processor for detecting a monochromatic plane wave signal in Gaussian noise is compared with a conventional beam-forming detector. Conditions are determined under which the optimum detector performs significantly better than the conventional detector. Conditions are also found under which the beamformer is itself near optimal.     

GPS C/N/sub 0/ estimation in the presence of interference and limited quantization levels

The carrier-to-noise density ratio (C/N₀) is considered an important parameter describing the GPS receiver performance. This paper compares the performance of two popular coarse-acquisition (C/A) C/N₀ algorithms appearing in literature: the variance summing method (VSM) (Psiaki et al., 2003, Psiaki, 2001), and the power ratio method (PRM) (Van Dierendonck, 1996, Sayre, 2003), in terms of their estimates in 1) additive white Gaussian noise (AWGN), 2) narrowband continuous wave interference (CWI), 3) their response to quantization and saturation effects, and their 4) dynamic range. The algorithms were implemented as a part of a software receiver. Two LI GPS data sets are examined; one was obtained from a GPS raw data collection setup, while the other was obtained from a GPS signal simulator. The collected set was stored with almost constant C/N₀ level while the simulated one contained variable C/N₀ levels. The effect of adding AWGN on the C/N₀ estimate was directly proportional with the noise power. The C/N₀ estimates suffered more when the CWI frequency was closer to the IF of the receiver. The PRM suffered from saturation at higher C/N₀ levels. The VSM showed good tracking at high C/N₀ levels and better immunity to limited quantization levels, while its C/N₀ estimate suffered from rapid fluctuations in power levels when sudden power steps occurred     

Optimal speckle reduction in polarimetric SAR imagery

Speckle is a major cause of degradation in synthetic aperture radar (SAR) imagery. With the availability of fully polarimetric SAR data, it is possible to use the three complex elements (HH, HV, VV) of the polarimetric scattering matrix to reduce speckle. The optimal method for combining the elements of the scattering matrix to minimize image speckle is derived, and the solution is shown to be a polarimetric whitening filter (PWF). A simulation of spatially correlated, K-distributed, fully polarimetric clutter is then used to compare the PWF with other, suboptimal speckle-reduction methods. Target detection performance of the PWF, span, and single-channel |HH|² detectors is compared with that of the optimal polarimetric detector (OPD). A novel, constant-false-alarm-rate (CFAR) detector (the adaptive PWF) is as a simple alternative to the OPD for detecting targets in clutter. This algorithm estimates the polarization covariance of the clutter, uses the covariance to construct the minimum-speckle image, and then tests for the presence of a target. An exact theoretical analysis of the adaptive PWF is presented; the algorithm is shown to have detection performance comparable with that of the OPD     

A Nominator-Selector Two-Stage Signal Detection Scheme

A two-stage approach to computationally efficient signal detection is to nominate candidate signal vectors using a suboptimal, but computationally cheap, detection scheme, and subsequently perform optimal detection on the candidates. Bounds on the detection performance of this approach are presented along with sufficient conditions for optimality. Examples are presented of an optimal and nearly optimal two-stage detection scheme where computational costs are drastically reduced relative to the singli-stage optimal detector.     

Parametric Rao Test for Multichannel Adaptive Signal Detection

The parametric Rao test for a multichannel adaptive signal detection problem is derived by modeling the disturbance signal as a multichannel autoregressive (AR) process. Interestingly, the parametric Rao test takes a form identical to that of the recently introduced parametric adaptive matched filter (PAMF) detector for space-time adaptive processing (STAP) in airborne surveillance radar systems and other similar applications. The equivalence offers new insights into the performance and implementation of the PAMF detector. Specifically, the Rao/PAMF detector is asymptotically (for large samples) a parametric generalized likelihood ratio test (GLRT), due to an asymptotic equivalence between the Rao test and the GLRT. The asymptotic distribution of the Rao test statistic is obtained in closed form, which follows an exponential distribution under the null hypothesis H ₀ and, respectively, a noncentral Chi-squared distribution with two degrees of freedom under the alternative hypothesis H ₁. The noncentrality parameter of the noncentral Chi-squared distribution is determined by the output signal-to-interference-plus-noise ratio (SINR) of a temporal whitening filter. Since the asymptotic distribution under H ₀ is independent of the unknown parameters, the Rao/PAMF asymptotically achieves constant false alarm rate (CFAR). Numerical results show that these results are accurate in predicting the performance of the parametric Rao/PAMF detector even with moderate data support.     

Numerical Results on the Convergence of Relative Efficiencies

This paper considers the detection of a known constant signal in an additive non-Gaussian noise under the assumptions of discrete time and statistically independent noise samples. The objective is to determine how large sample size must be before the easily computed asymptotic relative efficiency becomes a valid measure of performance. The exact small-sample error probabilities are calculated for a Neyman-Pearson optimal nonlinear detector consisting of a zeromemory nonlinearity followed by summation and threshold comparison. "Large-tailed" noise having a double exponential distribution is used as an example. The exact distribution of the test statistics for a linear detector and for the Neyman-Pearson optimal detector are calculated. Then the relative efficiency of the Neyman-Pearson optimal detector, as compared to a linear detector, is computed in order to study the rate of approach of the relative efficiency to its asymptotic value.     

Nonparametric rank detectors on quantized radar video signals

Uniform randomization of ties is required for defining distribution-free ranks of independent and identically distributed quantized samples. Formulas of rank probabilities are given and applied to radar detection under quantized video samples. For some detectors, and assuming Gaussian noise, the asymptotic loss L_∞(q) is calculated versus the normalized quantization step q, and the loss L(q) is estimated by Monte Carlo simulations. Both of these resulted in monotonic functions of q (0<q<1.1) that are independent of the other parameters. Furthermore, L(q)≈L_∞(q )⩽0.45 dB, as q<0.8. The quantization step q is normalized with respect to the noise standard deviation