WO2004006119A2 - Systeme de detection d'anomalies - Google Patents

Systeme de detection d'anomalies Download PDF

Info

Publication number
WO2004006119A2
WO2004006119A2 PCT/GB2003/002712 GB0302712W WO2004006119A2 WO 2004006119 A2 WO2004006119 A2 WO 2004006119A2 GB 0302712 W GB0302712 W GB 0302712W WO 2004006119 A2 WO2004006119 A2 WO 2004006119A2
Authority
WO
WIPO (PCT)
Prior art keywords
differences
input
signal
anomaly
measurements
Prior art date
Application number
PCT/GB2003/002712
Other languages
English (en)
Other versions
WO2004006119A3 (fr
Inventor
Timothy Raymond Field
Original Assignee
Qinetiq Limited
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Qinetiq Limited filed Critical Qinetiq Limited
Priority to AU2003240135A priority Critical patent/AU2003240135A1/en
Publication of WO2004006119A2 publication Critical patent/WO2004006119A2/fr
Publication of WO2004006119A3 publication Critical patent/WO2004006119A3/fr

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/28—Details of pulse systems
    • G01S7/285—Receivers
    • G01S7/292—Extracting wanted echo-signals
    • G01S7/2923—Extracting wanted echo-signals based on data belonging to a number of consecutive radar periods

Definitions

  • This invention relates to anomaly detection systems.
  • it relates to systems for the detection of anomalies in a medium having known or predictable statistical characteristics.
  • An anomaly in the context of this specification is an element within an observed field that does not behave in the same statistical manner as the surrounding medium.
  • Systems may be concerned with the detection of an anomaly on the surface of water or land, where the anomaly is an electromagnetic reflector, in which case the detection system may involve a radar, lidar, or other such electromagnetic observing equipment.
  • Systems may also be concerned with the propagation of signals through a heterogeneous media, such as laser light through the atmosphere, or acoustic signals propagating through gases or liquids.
  • Other systems may be concerned with financial time-series analysis, such as share price information, where it is advantageous to be able to separate anomalous behaviour from a well understood underlying process. Many of the signals involved in such systems may be modelled as a vector diffusion process. More information on this can be found in Oskendal, B. 1998, Stochastic Differential Equations - An Introduction with Applications, 5 th Edition. Springer
  • a filter may be formed by assuming that the overall shape of the clutter spectrum is a Gaussian, and estimating of the position of the central clutter frequency and the width of the clutter spectrum. The estimates are typically made by averaging several spectra from a region thought to contain only clutter. There are disadvantages with such schemes. Firstly, if the range of target frequencies falls within the clutter band, then the target will be suppressed after application, of the clutter filter. Secondly the estimate of the clutter spectrum may be contaminated with targets and / or that the average clutter spectrum may not adequately represent the clutter which is to be suppressed (i.e. there is a high degree of local variation in the clutter spectrum). Also, the clutter spectrum may change over the interval during which the several spectra are measured.
  • Another method of anomaly detection is to use a neural network, or self- organising system.
  • Such systems are able to discover structure in data, and hence potentially detect anomalies, with no prior knowledge of the statistics of the data.
  • One such system is the Stochastic Vector Quantiser, more details of which can be found in S. P. Luttrell, "Using Stochastic Vector Quantisers to characterise signal and noise subspaces", Proc. 5 th IMA Int. Conf. on Mathematics in Signal Processing, 2000.
  • a problem with this technique is that the neural network must be trained, which diverts time and resources from the actual detection process. Also, if a trained system is to be employed on inputs having different statistical properties, then the neural network must be retrained from scratch.
  • an anomaly detection system arranged to have as an input a signal comprising a plurality of measurements, and arranged to provide an output signal indicative of the presence of an anomaly, characterised in that the system is also arranged: a. to compute a set of differences between successive input measurements; b. to estimate or compute the expectation of the square of the modulus of the set of differences; c. to correlate the expectation estimated in b. with a function of the input signal to create a correlation signal; and d. to provide the output signal indicative of the presence of an anomaly, the output signal being generated from the correlation signal.
  • the invention provides the advantage that generation of an output indicative of an anomaly does not require an ensemble average of many input signals, nor does it require a training sequence. A single sequence of measurements is sufficient for the invention to provide a result as to the presence or otherwise of an anomaly.
  • the invention is applicable wherever the input data can be modelled as a stochastic process via a stochastic differential equation (SDE).
  • SDE stochastic differential equation
  • the volatility coefficient of the SDE which measures the amplitude of the local time fluctuation in the process, is particularly important since it is observable without recourse to an ensemble average. Identification of the volatility coefficient in the SDE enables a comparison to be made between the theoretically predicted and empirically observed volatilities.
  • the invention may be stated as follows: for any input quantity q t , l if no anomaly is present, and departs from 1 by some predetermined threshold amount to indicate an anomaly.
  • c(X,Y) is the correlation function, where X and Y are the empirically observed and theoretically predicted volatilities respectively.
  • the observed volatility term ⁇ q t ⁇ 2 comprises the set of differences between successive input signals, whilst with the theoretically predicted volatility term ?(q t ), the function ( must be deduced.
  • the " departure from unity of c to be regarded as indicative of an anomaly will be set according to system performance and system parameters such as desired false alarm rate etc.
  • Electromagnetic scattering from random media is typically r -distributed (as will be described later) in the intensity variable, and in such a -distributed domain the volatilities (as explained above) are closely correlated.
  • This feature is utilised in the current anomaly detection system when applied to electromagnetic input signals.
  • Such signals will generally be complex-valued.
  • the square of the observed volatility, ⁇ ⁇ 2 is approximately equal to the input signal z t , and so a departure from an approximately unity correlation between
  • Inputs from non-electromagnetic signals may be real-valued.
  • Such an example is a financial time series, S t .
  • St may, for example, comprise of company share price values overtime.
  • S r will not be r -distributed.
  • the function to be correlated with the set of differences will need to be determined from a model of the underlying financial process.
  • Stochastic volatility models describing financial data exist, and the invention can be applied to a suitable a 2 (St ) function derived from such a model, see e.g. the discussion of interest rate models for which the volatility function can take various forms according to the characteristics of the model, in Baxter, M. and Rennie, A. 1996 Financial Calculus - An introduction to derivative pricing, Cambridge University Press.
  • the system may be arranged to calculate the error between the intensity distribution of the input signal and a theoretical -distribution, and may provide an indication relating to the degree of match between the two. This may be used as an indication of the reliability of the result given by the anomaly detection system. Likewise, if the input signal is expected to follow some other intensity distribution, the intensity distribution of the actual input signal may be compared with the appropriate theoretical distribution and the comparison used to provide an indication of reliability.
  • the correlation function used to correlate the fluctuations with their prediction as some function of the input signal is preferably the standard statistical correlation function:
  • the input data may be sampled at intervals that need not be specified in advance.
  • the increment of a discrete-time sampled amplitude diffusion process, as utilised in the present invention, satisfies the equation
  • the sample interval is preferably small enough so that ⁇ t 112 » ⁇ t , which means that the volatility component, ⁇ t of equation 2 will dominate over the drift component ⁇ t , and so ⁇ 5 ⁇ , 2 » ⁇ 2 n 2 ⁇ t [+ higher order terms]
  • a root-sampling frequency or value in the case of the recast (B ⁇ t m ) » 1 , such as 10, 50, 100, 1000, 10000 or 100000.
  • a method of detecting an anomaly in an input signal comprising the steps: a. computing a set of differences between successive input measurements; b. estimating or computing the expectation of the square of the modulus of the set of differences; c. correlating the expectation estimated in b. with a function of the. input signal to create a correlation signal; and . d. comparing the correlation signal against a predetermined threshold, wherein an. anomaly is indicated should the correlation signal fall below the threshold.
  • Figure 1 diagrammatically illustrates a radar system in which may be implemented the current invention
  • Figure 2 shows a plot of unprocessed baseband radar intensity data comprising sea clutter and a single target
  • Figure 3 shows a graph of a theoretical K-distribution overlaid with the distribution of measured data
  • Figure 4 shows a graph of the correlation function with respect to range cell, for sea clutter data, according to the present invention
  • Figure 5 shows a graph of an alternate representation of the correlation function, such that the persistence over time of the anomaly can be seen
  • Figure 6 shows a graph of two signals input to the correlation function, the signals taken at a range where the correlation is greatest;
  • Figure 7 shows a graph of two signals input to the correlation function, the signals taken at a range where the correlation is least;
  • Figure 8 shows a graph of the permuted time representation of the smoothed data of Figure 6
  • Figure 9 shows a graph of the permuted time representation of the smoothed data of Figure 7
  • Figure 10 shows a block diagram of the steps performed on input data during the anomaly detection process
  • Illustrated in Figure 1 is a radar system able to transmit and receive electromagnetic waves.
  • the energy received will be a portion of the energy that has been transmitted towards a scene, and scattered from that scene.
  • the scene may typically comprise a target 1 that is floating on seawater 2.
  • the radar system comprises an antenna 3 that receives the scattered electromagnetic waves and passes them to a front end processing system 4.
  • the processing system 4 amplifies, mixes down and detects the energy, the detected energy comprising components reflected from the target itself, the waves and other elements of the seawater, and other stray signals and thermal noise.
  • the detected energy is digitised at this stage to produce a digital signal before being passed to the baseband signal processing section 5.
  • the current invention is implemented within this baseband signal processing section 5. After the digital signal has been processed in the signal processing section 5, it is passed to the display section 6 such that it may be observed by an operator.
  • Illustrated in Figure 2 is a plot of a baseband signal such as may be input to a baseband signal processing unit. It is raw, unprocessed data measured over a three second time interval and comprises a sequence of three thousand samples taken at 1ms intervals from a fixed viewpoint. Each sample contains information from 1024 range bins each 0.3 metres long. A tethered target is positioned at range bin 839, although this is not obvious from merely looking at Figure 2. Note that Figure 2 only shows the intensity - in reality, the data is complex, in-phase and quadrature-phase (l,Q) data that is used in the generation of an amplitude process of the form with t and x representing sample number and range respectively.
  • l,Q quadrature-phase
  • FIG. 3 shows the distribution of data taken from forward scattering measurements from a random phase screen (dotted line) overlaid with a best fit theoretical rC-distribution (solid line).
  • the random phase screen was created by producing a turbulent, rising airflow by means of natural convection from a heating element, and phase shifts imposed on laser radiation passing through the airflow used to generate the forward scattering measurement data.
  • the data shown in Figure 2 is in- phase and quadrature-phase (lt,Q t ) data (as described in Helmstrom, C W, "Statistical Theory of Signal Detection” p. 11 Pergamon, Oxford, 1960. that is used in the generation of a complex valued amplitude process of the form ⁇ t( ⁇ ) ⁇ lt( ⁇ )+iQt(x).
  • Temporal data from each range is taken in turn and processed in the following manner.
  • an estimate of the expectation (•) of J-5 ⁇ ,] is evaluated.
  • the estimation is an approximation of the expectation and is obtained by smoothing the raw values by averaging over a time window [t - ⁇ , t + A], where ⁇ is a smoothing parameter for variable t.
  • ⁇ is a smoothing parameter for variable t.
  • the samples may be ordered to give increased accuracy.
  • the input values are first permuted such that they are ordered according to their magnitude. Then, the same permutation is applied to the observed volatilities, and these volatilities are smoothed according to the above smoothing procedure. Finally, the permuted, smoothed samples are reordered back into their original time order.
  • Figure 4 shows a graph of the results of this process carried out at each range shown in Figure 2.
  • the smoothing parameter, ⁇ used for this example was set to 50 samples. This figure should be chosen such that the correlation in known non-anomalous ranges is as high as possible. It can be seen that most ranges have a high correlation, indicating that the input data from those ranges is substantially r -distributed, and hence is not behaving in an anomalous fashion. There is one clear null centered at range cell 839, indicative of a target present at this range.
  • the correlation plot may be input to a threshold circuit, and an output from this generated if the correlation falls below some predetermined figure.
  • Figure 5 again shows the correlation signal - shown as 1 -c for improved visualisation - but this time as a 3D plot, with each point on the time axis representing the average from a 1s window of data. The 20 points on the time axis thus represent 3 seconds of data.
  • the Figure 5 illustrates the persistence of the anomaly detection mechanism over time.
  • Figure 6 shows a comparison of two signals input to the correlation function, at a range where the value of the correlation is highest.
  • the solid line is ?(z t ) (i.e. z t whilst the dotted line represents the smoothed set of differences
  • Figure 7 shows a comparison of two signals as per Figure 6, but this time the signals are chosen so as to have a minimum correlation value.
  • the very low correlation value ( ⁇ 9.5%) is, in this embodiment, indicative of the presence of an anomaly. This correlation value is shown as the deep null in Figure 4
  • Figure 8 shows the permuted time representation of predicted and observed squared volatility for the process
  • the predicted values are permuted in ascending order; the required permutation is then also applied to the observed values, which are subsequently smoothed in the manner described above.
  • the range cell having the maximum correlation value was again chosen, and so the data in Figure 8 corresponds to the data shown in Figure 6.
  • Figure 9 is a similar plot to that shown in Figure 8, but this time the range cell having the minimum correlation value was chosen.
  • the data in Figure 9 therefore corresponds to the data shown in Figure 7.
  • the process carried out by the current invention is summarised in the block diagram of Figure 10. This shows the sequence of events occurring in the anomaly detection process specifically as applied to the radar embodiment.
  • Box (a) indicates the data input procedure.
  • the data should be sampled at a rate such that ⁇ t m » ⁇ t , as described above.
  • the sampling will typically be done using analogue to digital converters (ADCs), which act upon the data received from the radar front end, where the data is downconverted to baseband or to some other frequency at which it is convenient to process it digitally. Note that this sampling procedure is prior art.
  • ADCs analogue to digital converters
  • Box (b) shows the optional stage of checking the input data to see how closely it complies to a theoretical -distribution. If it is known that the input data should be K-distributed, then the data can be checked using the methods described above, and the resulting information used to provide a measure of confidence in the subsequent anomaly detection result. If the system were being used on data that was not K-distributed, such as financial data, then the check would be done on the appropriate theoretical distribution for that system.
  • Box (c) indicates the process of generating a set of differences between subsequent input readings.
  • the expectation of the square of the differences act as one of the inputs to the correlation stage indicated in box (d).
  • the other input is the theoretically predicted volatility for the inputs, c?( ⁇ Details of the calculation of the expectation are provided above. Note that a correlation value is generated for each range cell.
  • the volatility function ⁇ 2 (qD needs to be derived as appropriate.
  • volatility functions exist in many fields, such as the financial field, and those skilled in the relevant arts will be aware of how to go about deriving them " . '
  • the correlation value generated therein is input to a threshold circuit, which generates an output indicative of an anomaly if the threshold falls below a predetermined value. This is indicated by box (e).

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar Systems Or Details Thereof (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)

Abstract

La présente invention concerne un système de détection d'anomalies permettant de détecter des anomalies dans des signaux constitués de mesures successives, ledit système étant conçu pour : i) calculer un ensemble de différences entre des mesures d'entrée successives; ii) évaluer l'espérance du carré des modules de l'ensemble de différences; iii) corréler ladite espérance avec une fonction du signal d'entrée afin de créer un signal de corrélation; et iv) fournir une sortie indiquant la présence d'une anomalie, de préférence, par détermination du seuil du signal de corrélation. Dans la signification de la spécification, une anomalie est un réflecteur de signal qui ne réfléchit pas de la même manière statistique que son environnement, tel qu'un navire sur la mer. Dans d'autres modes de réalisation, les signaux sont principalement K-distribués, mais l'invention peut s'appliquer à des signaux présentant d'autres distributions.
PCT/GB2003/002712 2002-07-05 2003-06-24 Systeme de detection d'anomalies WO2004006119A2 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
AU2003240135A AU2003240135A1 (en) 2002-07-05 2003-06-24 Anomaly detection system

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
GB0215585A GB0215585D0 (en) 2002-07-05 2002-07-05 Anomaly detection system
GB0215585.1 2002-07-05

Publications (2)

Publication Number Publication Date
WO2004006119A2 true WO2004006119A2 (fr) 2004-01-15
WO2004006119A3 WO2004006119A3 (fr) 2004-04-15

Family

ID=9939907

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/GB2003/002712 WO2004006119A2 (fr) 2002-07-05 2003-06-24 Systeme de detection d'anomalies

Country Status (3)

Country Link
AU (1) AU2003240135A1 (fr)
GB (1) GB0215585D0 (fr)
WO (1) WO2004006119A2 (fr)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2483323C1 (ru) * 2011-12-22 2013-05-27 Закрытое акционерное общество "Морские комплексы и системы" Способ создания локационного изображения повышенной яркости и контрастности и устройство для его реализации
RU2503029C2 (ru) * 2011-12-28 2013-12-27 Игорь Федорович Шишкин Способ обнаружения аномалий на водной поверхности
RU2582073C2 (ru) * 2014-07-01 2016-04-20 Федеральное государственное казенное военное образовательное учреждение высшего профессионального образования "Военный учебно-научный центр Военно-Морского Флота "Военно-морская академия имени Адмирала Флота Советского Союза Н.Г. Кузнецова" Способ определения аномалий на морской поверхности неконтактным радиолокационным методом
DE102017220243B3 (de) 2017-11-14 2019-02-21 Continental Automotive Gmbh Verfahren und vorrichtung zum erkennen einer blockierung einer radarvorrichtung, fahrerassistenzsystem und fahrzeug
RU2686678C1 (ru) * 2018-01-31 2019-04-30 Акционерное общество "Военно-промышленная корпорация "Научно-производственное объединение машиностроения" Способ радиолокационного обзора морской поверхности и устройство для его осуществления

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4139847A (en) * 1976-06-16 1979-02-13 Japan Radio Company, Limited Automatic ground-clutter rejection in weather pulse radar system
US4532639A (en) * 1983-09-19 1985-07-30 Sperry Corporation CFAR receiver apparatus for detecting a signal in noise

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4139847A (en) * 1976-06-16 1979-02-13 Japan Radio Company, Limited Automatic ground-clutter rejection in weather pulse radar system
US4532639A (en) * 1983-09-19 1985-07-30 Sperry Corporation CFAR receiver apparatus for detecting a signal in noise

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
FARINA A ET AL: "Radar detection of targets: new theoretical findings and results by processing recorded live data" PROCEEDINGS OF THE 1996 CIE INTERNATIONAL CONFERENCE OF RADAR, BEIJING, CHINA, 8 - 10 October 1996, pages 5-12, XP002269072 *
FIELD T R, TOUGH R J A: "Diffusion processes in electromagnetic scattering generating K-distributed noise" LECTURES ON "STOCHASTIC DIFFERENTIAL EQUATIONS AND THEIR APPLICATION TO THE CHARACTERIZATION OF SEA CLUTTER", [Online] 7 - 8 October 2002, XP002269074 Adaptive Systems Laboratory, McMaster University, Canada Retrieved from the Internet: <URL:http://soma.crl.mcmaster.ca/ASLWeb/Re sources/data/TimothyField_October2002/K_di stribution_submission.pdf> [retrieved on 2004-02-04] *
FIELD T R, TOUGH R J A: "STOCHASTIC DYNAMICS OF THE SCATTERING AMPLITUDE GENERATING K-DISTRIBUTED NOISE" LECTURES ON "STOCHASTIC DIFFERENTIAL EQUATIONS AND THEIR APPLICATIONS TO THE CHARACTERIZATION OF SEA CLUTTER", [Online] 7 - 8 October 2002, XP002269075 Adaptive Systems Laboratory, McMaster University, Canada Retrieved from the Internet: <URL:http://soma.crl.mcmaster.ca/ASLWeb/Re sources/data/TimothyField_October2002/K_am plitude_submission.pdf> [retrieved on 2004-02-04] *
TOUGH R J A: "A Fokker-Planck description of K-distributed noise" JOURNAL OF PHYSICS A (MATHEMATICAL AND GENERAL), [Online] vol. 20, 1987, pages 551-567, XP002269073 Retrieved from the Internet: <URL:http://www.iop.org/EJ/article/0305-44 70/20/3/017/jav20i3p551.pdf> [retrieved on 2004-02-03] *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
RU2483323C1 (ru) * 2011-12-22 2013-05-27 Закрытое акционерное общество "Морские комплексы и системы" Способ создания локационного изображения повышенной яркости и контрастности и устройство для его реализации
RU2503029C2 (ru) * 2011-12-28 2013-12-27 Игорь Федорович Шишкин Способ обнаружения аномалий на водной поверхности
RU2582073C2 (ru) * 2014-07-01 2016-04-20 Федеральное государственное казенное военное образовательное учреждение высшего профессионального образования "Военный учебно-научный центр Военно-Морского Флота "Военно-морская академия имени Адмирала Флота Советского Союза Н.Г. Кузнецова" Способ определения аномалий на морской поверхности неконтактным радиолокационным методом
DE102017220243B3 (de) 2017-11-14 2019-02-21 Continental Automotive Gmbh Verfahren und vorrichtung zum erkennen einer blockierung einer radarvorrichtung, fahrerassistenzsystem und fahrzeug
RU2686678C1 (ru) * 2018-01-31 2019-04-30 Акционерное общество "Военно-промышленная корпорация "Научно-производственное объединение машиностроения" Способ радиолокационного обзора морской поверхности и устройство для его осуществления

Also Published As

Publication number Publication date
AU2003240135A8 (en) 2004-01-23
AU2003240135A1 (en) 2004-01-23
GB0215585D0 (en) 2002-08-14
WO2004006119A3 (fr) 2004-04-15

Similar Documents

Publication Publication Date Title
Watts Cell-averaging CFAR gain in spatially correlated K-distributed clutter
US7652617B2 (en) Radar microsensor for detection, tracking, and classification
Barbarossa et al. Detection and imaging of moving objects with synthetic aperture radar. Part 2: Joint time-frequency analysis by Wigner-Ville distribution
Rosenberg et al. Application of the Pareto plus noise distribution to medium grazing angle sea-clutter
Rosenberg et al. Application of the K+ Rayleigh distribution to high grazing angle sea-clutter
Yardim et al. Tracking of geoacoustic parameters using Kalman and particle filters
Rosenberg et al. Non-coherent radar detection performance in medium grazing angle X-band sea clutter
CN111610501A (zh) 对海雷达小目标检测方法
RU2704789C1 (ru) Способ адаптивной обработки сигналов в обзорных когерентно-импульсных радиолокационных станциях
Rosenberg et al. Coherent detection in medium grazing angle sea‐clutter
US6369749B1 (en) Adaptive control of the detection threshold of a binary integrator
Artyushenko et al. Nakagami distribution parameters comparatively estimated by the moment and maximum likelihood methods
Rosenberg et al. Model based coherent detection in medium grazing angle sea-clutter
WO2004006119A2 (fr) Systeme de detection d&#39;anomalies
Aloisio et al. Optimum detection of moderately fluctuating radar targets
Tohidi et al. Compressive sensing MTI processing in distributed MIMO radars
Watts et al. Coherent radar performance in sea clutter
Rosenberg et al. Performance analysis of Pareto CFAR detectors
Prokopenko et al. Implementation of Adaptive Algorithms in the Task of MTDI Filtration
Conte et al. Incoherent radar detection in compound-Gaussian clutter
Watts Modelling of coherent detectors in sea clutter
Rosenberg Coherent detection with non-stationary high grazing angle X-band sea-clutter
Ing et al. Parametric texture estimation and prediction using measured sea clutter data
Doukovska Constant false alarm rate detectors in intensive noise environment conditions
Kumar et al. Classification of radar returns using Wigner-Ville distribution

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A2

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A2

Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
122 Ep: pct application non-entry in european phase
NENP Non-entry into the national phase in:

Ref country code: JP

WWW Wipo information: withdrawn in national office

Country of ref document: JP