A Posteriori Probability Distances Between Finite-Alphabet Hidden Markov Models

In this correspondence, we consider a probability distance problem for a class of hidden Markov models (HMMs). The notion of conditional relative entropy between conditional probability measures is introduced as an a posteriori probability distance which can be used to measure the discrepancy between hidden Markov models when a realized observation sequence is observed. Using a measure change technique, we derive a representation for conditional relative entropy in terms of the parameters of the HMMs and conditional expectations given measurements. With this representation, we show that this distance can be calculated using an information state approach     

Properties of cross-entropy minimization

The principle of minimum cross-entropy (minimum directed divergence, minimum discrimination information) is a general method of inference about an unknown probability density when there exists a prior estimate of the density and new information in the form of constraints on expected values. Various fundamental properties of cross-entropy minimization are proven and collected in one place. Cross-entropy's well-known properties as an information measure are extended and strengthened when one of the densities involved is the result of cross-entropy minimization. The interplay between properties of cross-entropy minimization as an inference procedure and properties of cross-entropy as an information measure is pointed out. Examples are included and general analytic and computational methods of finding minimum cross-entropy probability densities are discussed.     

Cycle slipping in an optical communication system employing subcarrier angle modulation

The output of an ideal photodetector that is modeled as a doubly stochastic Poisson process is used to excite a phase-locked loop. The rate of this Poisson process contains an angle-modulated subcarrier process that is modeled as a Gauss-Markov process. A measure transformation involving a martingale translation is applied to the suboptimal filter used to estimate the phase. Cycle slipping is then discussed with respect to the evolution of the phase error process under the new measure. The approach is in the context of a first-passage problem and uses a Kramers-Moyal expansion of the equation satisfied by the mean slip time. Various approximations to the jump rate and size are assumed, and a perturbation analysis results in asymptotic representations of the corresponding solutions. Comparisons with solutions obtained from so-called diffusion approximations are also included, as well as a specific example to illustrate the procedure.     

融合局部加权余弦与稀疏表示的目标跟踪算法

针对目标跟踪算法的鲁棒性难题,在粒子滤波框架下提出基于联合模型的目标跟踪算法。首先,由局部加权余弦相似对目标模板和候选目标进行匹配,其中的局部加权算法增加了未受遮挡、形变等影响的候选目标的权重;其次,通过对目标区域局部图像块稀疏编码来表示目标观测模型,其中字典不进行更新,重建误差的构建考虑了局部图像块之间的空间布局;最后,利用最大后验概率估计目标状态。联合模型将目标的当前状态和原始状态都考虑在内,提高了观测模型的可靠性。实验结果表明,该算法具有较强的鲁棒性。     

基于粒子滤波的协作式空时联合频谱检测方法

认知无线电系统中,通过频谱检测所获得的主用户位置信息和信道使用状态信息分别反映空间域和时间域的频谱使用情况,其准确性关系到次用户伺机频谱接入的效率。本文研究空间和时间联合频谱检测问题,将其建模为主用户位置参数和信道使用状态的联合估计问题,提出了一种基于粒子滤波的协作式空时联合频谱检测方法,主要思想是采用离散随机测度逼近位置参数和信道状态的联合后验概率分布函数,根据该分布估计出主用户位置和信道状态。同时,该方法结合由估计得到的空间和时间信息,自适应选择靠近主用户且对称分布的次用户参与协作检测,增加各次用户提供信息的有效性,从而提高估计的准确性;最后,通过仿真验证了方法的有效性。      

Regular jump processes and their information processing

A class of regular jump processes (RJP's) is introduced. An RJP is described in terms of the intensity function of its associated stochastic point process and the state-transition density of its embedded random-state sequence. Expressions for the joint occurrence statistics of these processes are derived. Assuming that an information stochastic process causally modulates an observed RJP, we obtain the joint occurrence statistics of the resulting compound jump processes. We show the latter to incorporate appropriately the causal MMSE estimate of the conditional intensities and state-transition functions. The results are used to derive a general likelihood-ratio formula for information processing of RJP's. A separation is observed between the likelihood processor of the point process associated with the observed RJP and the processor associated with the embedded stochastic state sequence. Considering the detection of RJP's with uncertain (statistically known) probability measures, we obtain the optimal Bayes receiver as the appropriate compound likelihood processor and thus exhibit separation between the detection and filtering operations.     

Characteristics and properties of signals space built on generalized Boolean algebra with measure

The characteristics of an arbitrary pair of random processes, which are invariant with respect to the group of random process transformations, are considered. By applying an apparatus of Boolean algebra with measure to the introduced characteristics of random processes, a notion of metric signals space with axiomatic introduction of information quantity measure, which allows to estimate information ratios between random signals and their instantaneous values, is formulated.     

基于子孔径参数估计的双基地ISAR图像融合方法研究

双基地逆合成孔径雷达(Bistatic ISAR)能够同时获取两个视角下的目标雷达回波,将分别所成的图像进行融合处理可以获得质量更高的ISAR图像以提供更丰富的目标信息,但由双基地角带来的目标像尺寸形状不一致等问题将造成融合困难。该文针对转台模型目标,提出一种基于子孔径参数估计的双基地ISAR图像融合方法,利用孤立强散射点在不同子孔径下的多普勒差值估计出目标的转动角速度和半双基地角,再将不同视角下的数据映射到同一个坐标平面中进行融合处理,得到效果更好的图像,最后利用仿真实验结果验证了该方法的有效性。     

基于有理数倍采样的异步数据融合算法研究

本文研究了一类具有不同采样率的分布式多传感器动态系统的数据融合问题,针对一类采样率呈有理数倍关系的动态系统,提出一种基于多源异步采样数据的新融合算法.新算法首先是将来自各个传感器的测量值在融合中心的坐标系中和时钟下进行映射统一;其次,以对目标状态下一时刻的预测值与目标在该时刻状态的估计值之差为基础,建立起描述该融合周期内各个观测点处的目标状态向量之间的动态模型;然后,以该时刻目标状态基于全局信息的估计值为条件,结合建立的新模型和传统的Kalman滤波器,利用本周期内按序到达的各传感器观测值,依次对各个观测点处目标的状态进行估计和更新;最后,在顺序得到本周期内各个观测点处目标估计值的同时,也将获得下一时刻目标状态基于全局信息的估计值或预测估计值.文中在给出新算法基本思想的同时,也较为详细地对融合算法进行了推导,并通过计算机仿真的方法,将新算法与基于时间校准的算法在估计精确度上进行了比较,从而验证了新算法的有效性.     

基于置信区域内多级动态层表达的类贯序蒙特卡洛视觉跟踪方法

视觉跟踪是智能监控、机器人和视觉导航等领域的核心技术.针对现有类贯序蒙特卡洛跟踪方法复杂度高、实时性差的问题,本文提出了一种融合置信区域内多级动态层表达的跟踪方法,采用更加可靠、有效的粒子模拟状态后验概率.该方法利用检测模块得到目标可能出现的置信区域,根据真实目标尺寸给出一种粒子采样策略,每个粒子代表一级动态层表达,并为每个粒子建立双层运动模型;构建Mean-Shift分块观测模型以引入空间和外观信息,同时根据子块的匹配程度计算粒子权值、检测目标遮挡状态并提出模型更新策略.在公开视频序列上同经典粒子滤波和Mean-Shift等算法的实验对比结果证明了本文算法具有较优的跟踪准确度和实时性.     

基于二阶隐马尔可夫模型的文本信息抽取

隐马尔可夫模型是文本信息抽取的重要方法之一.在一阶隐马尔可夫模型中,假设状态转移概率和观察值输出概率仅依赖于模型当前的状态,一定程度降低了信息抽取的精确度.而二阶隐马尔可夫模型合理地考虑了概率和模型历史状态的关联性,对错误信息有更强的识别能力.提出了基于二阶隐马尔可夫模型的文本信息抽取算法;分析了二阶隐马尔可夫模型在文本信息抽取中的有效性;仿真实验表明,新的算法比基于一阶隐马尔可夫模型的算法具有更高的抽取精确度.     

基于五粒子团簇态实现四粒子团簇态的概率隐形传态

为实现更经济的四粒子团簇态的概率隐形传输,提出了一种利用五粒子团簇态为量子信道的隐形传态方案.基于解纠缠-隐形传输-重构纠缠方法,发送者在发送信息之前对信息态进行一次控制非门操作和投影测量,将解纠缠的信息发送给接收者.在这个过程中发送者需要进行2次Bell基联合测量和1次单粒子测量,接收者根据发送者的测量结果进行相应操作,得出解纠缠的信息态.引入辅助粒子重构原始的未知信息态,信息传输成功的概率为4|a|2,提出的方案可以很好地应对一般窃听方式.     

Robust Neural-Network-Based Data Association and Multiple Model-Based Tracking of Multiple Point Targets

Data association and model selection are important factors for tracking multiple targets in a dense clutter environment without using a priori information about the target dynamic. We propose a neural-network-based tracking algorithm, incorporating a interacting multiple model and show that it is possible to track both maneuvering and nonmaneuvering targets simultaneously in the presence of dense clutter. Moreover, it can be used for real-time application. The proposed method overcomes the problem of data association by using the method of expectation maximization and Hopfield network to evaluate assignment weights. All validated observations are used to update the target state. In the proposed approach, a probability density function (pdf) of an observed data, given target state and observation association, is treated as a mixture pdf. This allows to combine the likelihood of an observation due to each model, and the association process is defined to incorporate an interacting multiple model, and consequently, it is possible to track any arbitrary trajectory     

A global scattering center representation of complex targets usingthe shooting and bouncing ray technique

We present a methodology to extract a global scattering center model a complex target. Using the shooting and bouncing ray technique, we first extract the three-dimensional (3-D) scattering center representations of the target at various aspect angles. Next the correspondence between scattering centers across angles are established by utilizing 3-D location information. The data are grouped based on 3-D scattering center locations and are organized in the form of angular visibility maps. Such data organization gives good insights into target physics and feature stability. Furthermore, we define a stability measure to quantitatively rank order the scattering centers based on their angular stability. Finally, we demonstrate the usefulness of such stable features by carrying out a five-target template-based matching experiment to estimate the angular pose of an unknown target     

The performance of trellis-coded MDPSK with multiple symboldetection

The idea of using a multiple (more than two) symbol observation interval to improve error probability performance is applied to differential detection of trellis-coded multiple phase-shift keying (MPSK) over an additive white Gaussian noise (AWGN) channels. An equivalent Euclidean distance measure per trellis branch is determined for this detection scheme. This is used to define an augmented (larger multiplicity) trellis code whose distance measure is the conventional squared Euclidean distance typical of conventional trellis-coded modulation on the AWGN. Such an augmented multiple trellis code is a convenient mathematical tool for simplifying the analysis. Results are obtained by a combination of analysis and computer simulation. It is shown that only a slight increase (e.g. one symbol) in the length of the observation interval will provide a significant improvement in bit error probability performance     

An upper bound on average estimation error in nonlinear systems

An upper bound is obtained on the probability density of the estimate of the parametermwhen a nonlinear functions(t, m)is transmitted over a channel that adds Gaussian noise, and maximum likelihood or maximum a posteriori estimation is used. If this bound is integrated with a loss function, an upper bound on the average error is obtained. Nonlinear (below threshold) effects are included. The problem is viewed in a Euclidean space. Evaluation of the probability density can be reduced to integrating the probability density of the observation over part of a hyperplane. By bounding the integrand, and using a larger part of the hyperplane, an upper bound is obtained. The resulting bound on mean-square error is quite close for the cases calculated.     

Filtering and prediction from uncertain observations with correlated signal and noise via mixture approximations

A recursive suboptimal filtering and prediction algorithm is designed to estimate Gaussian signals from uncertain observations, when the variables describing the uncertainty are independent, and the signal and observation white noise are correlated. The derivation is based on successive approximations of mixtures of normal distributions which, in turn, provide approximations of the conditional distributions of the signal given the observations. The proposed estimators, which are nonlinear functions of the observations, are then obtained as the expectation of these approximate conditional distributions. The algorithm does not require the whole knowledge of the state-space model generating the signal, but only covariance information of the signal and the observation noise, as well as the probability that the signal exists in the observed values.     

Analysis and optimization of adaptive multicopy transmission ARQprotocols for time-varying channels

This paper outlines an efficient method to concurrently optimize a multiplicity of design variables for continuous selective-repeat (SR) and go-back-N (GBN) automatic repeat request (ARQ) strategies, both in noiseless and noisy feedback channels. For these ARQ protocols, we adapt either the number of identical message blocks sent in each transmission (in the case of GBN scheme) or the number of copies of a block retransmitted to handle a NACKed codeword (for the SR protocol) dynamically to the estimated channel condition. The channel state information is obtained by counting the contiguous acknowledgment (ACK or NACK) messages. Exploiting the asymptotic properties of the steady state probability expressions, we show analytically that the optimum solution indeed lies in the infinite space. Subsequently, a simple expression to estimate the suboptimal design parameters is suggested. Our approach of minimizing the mean-square error function yields to a quantitative study of the appropriateness of the selected parameters. Exact analytical expressions that allows us to compute the throughput crossover probability between any two arbitrary multicopy transmission modes are derived. The results provide fundamental insights into how these key parameters interact and determine the system performance     

Vision-Based Target Tracking and Surveillance With Robust Set-Valued State Estimation

Tracking a target from a video stream (or a sequence of image frames) involves nonlinear measurements in Cartesian coordinates. However, the target dynamics, modeled in Cartesian coordinates, result in a linear system. We present a robust linear filter based on an analytical nonlinear to linear measurement conversion algorithm. Using ideas from robust control theory, a rigorous theoretical analysis is given which guarantees that the state estimation error for the filter is bounded, i.e., a measure against filter divergence is obtained. In fact, an ellipsoidal set-valued estimate is obtained which is guaranteed to contain the true target location with an arbitrarily high probability. The algorithm is particularly suited to visual surveillance and tracking applications involving targets moving on a plane.