Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378815
A. Balestrino, A. Caiti, E. Crisostomi
The paper introduces a new estimation algorithm that blends together particle filtering techniques and set-membership theory to provide more complete and reliable state estimates. The algorithm is applied to linear time-discrete dynamic systems where the process and the measurement noises are combined with model uncertainties through ellipsoidal constraints; the algorithm however can be extended as well to mild non linear systems by replacing nonlinearities with uncertainties in the system matrices. Each step of the proposed estimation method is described in detail, and some simulation results are provided to show the behaviour of the algorithm.
{"title":"PP algorithm for Particle Filtering within Ellipsoidal Regions","authors":"A. Balestrino, A. Caiti, E. Crisostomi","doi":"10.1109/NSSPW.2006.4378815","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378815","url":null,"abstract":"The paper introduces a new estimation algorithm that blends together particle filtering techniques and set-membership theory to provide more complete and reliable state estimates. The algorithm is applied to linear time-discrete dynamic systems where the process and the measurement noises are combined with model uncertainties through ellipsoidal constraints; the algorithm however can be extended as well to mild non linear systems by replacing nonlinearities with uncertainties in the system matrices. Each step of the proposed estimation method is described in detail, and some simulation results are provided to show the behaviour of the algorithm.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"53 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129191190","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378860
V. Šmídl, A. Quinn
The Variational Bayes (VB) approach is used as a one-step approximation for Bayesian filtering. It requires the availability of moments of the free-form distributional optimizers. The latter may have intractable functional forms. In this contribution, we replace these by appropriate fixed-form distributions yielding the required moments. We address two scenarios of this Restricted VB (RVB) approximation. For the first scenario, an application in identification of HMMs is given. Close relationship of the second scenario to Rao-Blackwellized particle filtering is discussed and their performance is illustrated on a simple non-linear model.
{"title":"The Restricted Variational Bayes Approximation in Bayesian Filtering","authors":"V. Šmídl, A. Quinn","doi":"10.1109/NSSPW.2006.4378860","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378860","url":null,"abstract":"The Variational Bayes (VB) approach is used as a one-step approximation for Bayesian filtering. It requires the availability of moments of the free-form distributional optimizers. The latter may have intractable functional forms. In this contribution, we replace these by appropriate fixed-form distributions yielding the required moments. We address two scenarios of this Restricted VB (RVB) approximation. For the first scenario, an application in identification of HMMs is given. Close relationship of the second scenario to Rao-Blackwellized particle filtering is discussed and their performance is illustrated on a simple non-linear model.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132817787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378814
A. Runnalls
The IGMARP Data Fusion Algorithm Andrew R. Runnalls University of Kent Computing Laboratory Technical Report 05-07 IGMARP (Iterative Gaussian Mixture Approximation of the Reduced-Dimension Posterior) is a data fusion algorithm for handling non-linear measurements, particularly ambiguous measurements (i.e. measurements for which the likelihood function may be multimodal), in conjunction with a linear or linearisable system model. It is particularly well suited to system models of high dimensionality, and applications where it is desired to interoperate with existing approaches using a Kalman Filter or multi-hypothesis Kalman Filter. The algorithm was developed under sponsorship from QinetiQ Ltd over the period 2001-5 as a means of integrating data from terrain-referenced navigation systems into a multiway integrated navigation solution also comprising an inertial navigation system (INS) and GPS. An example of a terrain-referenced navigation system is terrain-contour navigation (TCN), in which an air vehicle uses a radio altimeter or similar sensor to take measurements of the height above sea level of the terrain being overflown. The paper describes the mathematical foundations of the algorithm, and illustrates its application to an integrated TCN/INS system. Sec. 2 introduces the motivating application, TCN. Sec. 3 reviews the measurement update equations for the multi-hypothesis Kalman filter (MHKF), which represent an application of Bayes' Theorem to the case in which the prior distribution is a Gaussian mixture, and the likelihood function also has the form of a (slightly generalised) Gaussian mixture. Sec. 4 then discusses how the likelihood function can be computed for TCN, and gives the flavour of the resulting functions, which are by no means of a Gaussian mixture form; this motivates Sec. 5, which discusses how the MHKF approach can be adapted to handle more general likelihood functions, and introduces the key theorems on which the IGMARP method depends. Then Sec. 6 describes the algorithm itself, and Sec. 7 illustrates the results of applying the algorithm to TCN/INS flight data. Finally Sec. 8 discusses conclusions and possible further work.
{"title":"The IGMARP Data Fusion Algorithm","authors":"A. Runnalls","doi":"10.1109/NSSPW.2006.4378814","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378814","url":null,"abstract":"The IGMARP Data Fusion Algorithm Andrew R. Runnalls University of Kent Computing Laboratory Technical Report 05-07 IGMARP (Iterative Gaussian Mixture Approximation of the Reduced-Dimension Posterior) is a data fusion algorithm for handling non-linear measurements, particularly ambiguous measurements (i.e. measurements for which the likelihood function may be multimodal), in conjunction with a linear or linearisable system model. It is particularly well suited to system models of high dimensionality, and applications where it is desired to interoperate with existing approaches using a Kalman Filter or multi-hypothesis Kalman Filter. The algorithm was developed under sponsorship from QinetiQ Ltd over the period 2001-5 as a means of integrating data from terrain-referenced navigation systems into a multiway integrated navigation solution also comprising an inertial navigation system (INS) and GPS. An example of a terrain-referenced navigation system is terrain-contour navigation (TCN), in which an air vehicle uses a radio altimeter or similar sensor to take measurements of the height above sea level of the terrain being overflown. The paper describes the mathematical foundations of the algorithm, and illustrates its application to an integrated TCN/INS system. Sec. 2 introduces the motivating application, TCN. Sec. 3 reviews the measurement update equations for the multi-hypothesis Kalman filter (MHKF), which represent an application of Bayes' Theorem to the case in which the prior distribution is a Gaussian mixture, and the likelihood function also has the form of a (slightly generalised) Gaussian mixture. Sec. 4 then discusses how the likelihood function can be computed for TCN, and gives the flavour of the resulting functions, which are by no means of a Gaussian mixture form; this motivates Sec. 5, which discusses how the MHKF approach can be adapted to handle more general likelihood functions, and introduces the key theorems on which the IGMARP method depends. Then Sec. 6 describes the algorithm itself, and Sec. 7 illustrates the results of applying the algorithm to TCN/INS flight data. Finally Sec. 8 discusses conclusions and possible further work.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"140 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133371758","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378842
U. Ramdaras, F. Absil
This paper presents a novel sensor selection algorithm for target tracking, based on the Modified Riccati Equation (MRE). The MRE provides an upper bound of the Cramér-Rao lower bound (CRLB) and is easily calculated. Using the MRE, it is possible to include sensors with a probability of detection Pd ≪ 1. State estimation is done with a modified Particle Filter (PF), taking into account missed detections. The performance of the MRE sensor selection scheme is studied for single and multiple steps ahead, and, for the case of Pd = 1, compared with other methods.
{"title":"Networks of Maritime Radar Systems: Sensor Selection Algorithm for PD ≪ 1 Based on the Modified Riccati Equation","authors":"U. Ramdaras, F. Absil","doi":"10.1109/NSSPW.2006.4378842","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378842","url":null,"abstract":"This paper presents a novel sensor selection algorithm for target tracking, based on the Modified Riccati Equation (MRE). The MRE provides an upper bound of the Cramér-Rao lower bound (CRLB) and is easily calculated. Using the MRE, it is possible to include sensors with a probability of detection Pd ≪ 1. State estimation is done with a modified Particle Filter (PF), taking into account missed detections. The performance of the MRE sensor selection scheme is studied for single and multiple steps ahead, and, for the case of Pd = 1, compared with other methods.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"58 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117152534","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Let x = {Xn}n IN be a hidden process, y = {yn}n IN an observed process and r = {rn}n IN some auxiliary process. We assume that t = {tn}n IN with tn = (xn, rn, yn-1) is a (Triplet) Markov Chain (TMC). TMC are more general than Hidden Markov Chains (HMC) and yet enable the development of efficient restoration and parameter estimation algorithms. This paper is devoted to Bayesian smoothing algorithms for TMC. We first propose twelve algorithms for general TMC. In the Gaussian case, they reduce to a set of algorithms which includes, among other solutions, extensions to TMC of classical Kalman-like smoothing algorithms such as the RTS algorithms, the Two-Filter algorithm or the Bryson and Frazier algorithm. We finally propose particle filtering (PF) approximations for the general case.
{"title":"Exact and Approximate Bayesian Smoothing Algorithms in Partially Observed Markov Chains","authors":"B. Ait‐El‐Fquih, F. Desbouvries","doi":"10.1063/1.2423292","DOIUrl":"https://doi.org/10.1063/1.2423292","url":null,"abstract":"Let x = {X<inf>n</inf>}<inf>n IN</inf> be a hidden process, y = {y<inf>n</inf>}<inf>n IN</inf> an observed process and r = {r<inf>n</inf>}<inf>n IN</inf> some auxiliary process. We assume that t = {t<inf>n</inf>}<inf>n IN</inf> with t<inf>n</inf> = (x<inf>n</inf>, r<inf>n</inf>, y<inf>n-1</inf>) is a (Triplet) Markov Chain (TMC). TMC are more general than Hidden Markov Chains (HMC) and yet enable the development of efficient restoration and parameter estimation algorithms. This paper is devoted to Bayesian smoothing algorithms for TMC. We first propose twelve algorithms for general TMC. In the Gaussian case, they reduce to a set of algorithms which includes, among other solutions, extensions to TMC of classical Kalman-like smoothing algorithms such as the RTS algorithms, the Two-Filter algorithm or the Bryson and Frazier algorithm. We finally propose particle filtering (PF) approximations for the general case.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127174270","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378810
R. Pradeepa, G. V. Anand
Non-Gaussianity of signals/noise often results in significant performance degradation for systems, which are designed using the Gaussian assumption. So non-Gaussian signals/noise require a different modelling and processing approach. In this paper, we discuss a new Bayesian estimation technique for non-Gaussian signals corrupted by colored non Gaussian noise. The method is based on using zero mean finite Gaussian Mixture Models (GMMs) for signal and noise. The estimation is done using an adaptive non-causal nonlinear filtering technique. The method involves deriving an estimator in terms of the GMM parameters, which are in turn estimated using the EM algorithm. The proposed filter is of finite length and offers computational feasibility. The simulations show that the proposed method gives a significant improvement compared to the linear filter for a wide variety of noise conditions, including impulsive noise. We also claim that the estimation of signal using the correlation with past and future samples leads to reduced mean squared error as compared to signal estimation based on past samples only.
{"title":"Estimation of Signals in Colored Non Gaussian Noise Based on Gaussian Mixture Models","authors":"R. Pradeepa, G. V. Anand","doi":"10.1109/NSSPW.2006.4378810","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378810","url":null,"abstract":"Non-Gaussianity of signals/noise often results in significant performance degradation for systems, which are designed using the Gaussian assumption. So non-Gaussian signals/noise require a different modelling and processing approach. In this paper, we discuss a new Bayesian estimation technique for non-Gaussian signals corrupted by colored non Gaussian noise. The method is based on using zero mean finite Gaussian Mixture Models (GMMs) for signal and noise. The estimation is done using an adaptive non-causal nonlinear filtering technique. The method involves deriving an estimator in terms of the GMM parameters, which are in turn estimated using the EM algorithm. The proposed filter is of finite length and offers computational feasibility. The simulations show that the proposed method gives a significant improvement compared to the linear filter for a wide variety of noise conditions, including impulsive noise. We also claim that the estimation of signal using the correlation with past and future samples leads to reduced mean squared error as compared to signal estimation based on past samples only.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128337794","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378843
J. Guillet, F. LeGland
Localization, navigation and tracking form a special application domain of Bayesian filtering, where the position and velocity of a mobile (and possibly additional hyper-parameters) should be estimated based on (i) a prior model for the possible displacement of the mobile, (ii) noisy measurements provided by a sensor, and (iii) a georeferenced information source (digital map, reference data base, etc.), providing for each spatial position an estimate of the quantity measured by the sensor. For example in terrain-aided navigation (TAN) a radio-altimeter combined with an inertial navigation system (INS) provides an estimation of the terrain height below the platform, which can be correlated with the terrain height at each horizontal position, as read on a digital map. In wireless communications, the signal power received by the mobile from an access point (WiFi) or from a base station (GSM, UMTS) and measured by the mobile itself, can be correlated with another estimation of the signal power received at each spatial position, as read on a digital attenuation map or from a reference data base. Values read on a digital map are usually subject to errors which are in general spatially correlated and modeled as Gaussian random fields, with a known correlation function. This results in a temporal correlation of measurement noises, which should be accounted for in evaluating the likelihood function, an essential step in the derivation of the equation for the Bayesian filter.
{"title":"Using Noisy Georeferenced Information Sources for Navigation and Tracking","authors":"J. Guillet, F. LeGland","doi":"10.1109/NSSPW.2006.4378843","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378843","url":null,"abstract":"Localization, navigation and tracking form a special application domain of Bayesian filtering, where the position and velocity of a mobile (and possibly additional hyper-parameters) should be estimated based on (i) a prior model for the possible displacement of the mobile, (ii) noisy measurements provided by a sensor, and (iii) a georeferenced information source (digital map, reference data base, etc.), providing for each spatial position an estimate of the quantity measured by the sensor. For example in terrain-aided navigation (TAN) a radio-altimeter combined with an inertial navigation system (INS) provides an estimation of the terrain height below the platform, which can be correlated with the terrain height at each horizontal position, as read on a digital map. In wireless communications, the signal power received by the mobile from an access point (WiFi) or from a base station (GSM, UMTS) and measured by the mobile itself, can be correlated with another estimation of the signal power received at each spatial position, as read on a digital attenuation map or from a reference data base. Values read on a digital map are usually subject to errors which are in general spatially correlated and modeled as Gaussian random fields, with a known correlation function. This results in a temporal correlation of measurement noises, which should be accounted for in evaluating the likelihood function, an essential step in the derivation of the equation for the Bayesian filter.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"17 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128210474","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378844
S. Julier, T. Bailey, J. Uhlmann
In this paper we investigate the use of Exponential Mixture Densities (EMDs) as suboptimal update rules for distributed data fusion. We show that EMDs have a pointwise bound "from below" on the minimum value of the probability distribution. However, the distributions are not bounded from above and thus can be interpreted as a fusion operation.
{"title":"Using Exponential Mixture Models for Suboptimal Distributed Data Fusion","authors":"S. Julier, T. Bailey, J. Uhlmann","doi":"10.1109/NSSPW.2006.4378844","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378844","url":null,"abstract":"In this paper we investigate the use of Exponential Mixture Densities (EMDs) as suboptimal update rules for distributed data fusion. We show that EMDs have a pointwise bound \"from below\" on the minimum value of the probability distribution. However, the distributions are not bounded from above and thus can be interpreted as a fusion operation.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"95 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132591578","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378832
M. Jaward, D. Bull, N. Canagarajah
Nonlinear distributed tracking for a single target is addressed in this paper. This problem consists of tracking a target of interest while moving the sensors to `best' positions according to an critera appropriate for the problem. Both target tracking and manoeuvring of sensors are carried out jointly using a novel Sequential Monte Carlo technique. The proposed technique is illustrated using a bearing-only problem and simulations are used to compare the performance of the proposed technique with distributed tracking using fixed sensors.
{"title":"Distributed Tracking with Sequential Monte Carlo Methods for Manoeuvrable Sensors","authors":"M. Jaward, D. Bull, N. Canagarajah","doi":"10.1109/NSSPW.2006.4378832","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378832","url":null,"abstract":"Nonlinear distributed tracking for a single target is addressed in this paper. This problem consists of tracking a target of interest while moving the sensors to `best' positions according to an critera appropriate for the problem. Both target tracking and manoeuvring of sensors are carried out jointly using a novel Sequential Monte Carlo technique. The proposed technique is illustrated using a bearing-only problem and simulations are used to compare the performance of the proposed technique with distributed tracking using fixed sensors.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"72 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133887473","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2006-09-01DOI: 10.1109/NSSPW.2006.4378862
C. Kreucher, A. Hero
Surveillance for multi-target detection, identification and tracking is one of the natural problem domains in which particle filtering approaches have been gainfully applied. Sequential importance sampling is used to generate and update estimates of the joint multi-target probability density for the number of targets, their dynamical model, and their state vector. In many cases there are a large number of degrees of freedom in sensor deployment, e.g., choice of waveform or modality. This gives rise to a resource allocation problem that can be formulated as determining an optimal policy for a partially observable Markov decision process (POMDP). In this paper we summarize approaches to solving this problem which involve using particle filtering to estimate both posterior state probabilities and the expected reward for both myopic and multistage policies.
{"title":"Monte Carlo Methods for Sensor Management in Target Tracking","authors":"C. Kreucher, A. Hero","doi":"10.1109/NSSPW.2006.4378862","DOIUrl":"https://doi.org/10.1109/NSSPW.2006.4378862","url":null,"abstract":"Surveillance for multi-target detection, identification and tracking is one of the natural problem domains in which particle filtering approaches have been gainfully applied. Sequential importance sampling is used to generate and update estimates of the joint multi-target probability density for the number of targets, their dynamical model, and their state vector. In many cases there are a large number of degrees of freedom in sensor deployment, e.g., choice of waveform or modality. This gives rise to a resource allocation problem that can be formulated as determining an optimal policy for a partially observable Markov decision process (POMDP). In this paper we summarize approaches to solving this problem which involve using particle filtering to estimate both posterior state probabilities and the expected reward for both myopic and multistage policies.","PeriodicalId":388611,"journal":{"name":"2006 IEEE Nonlinear Statistical Signal Processing Workshop","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2006-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127697483","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：