Robert R. Tenney

Detection with distributed sensors

The extension of classical detection theory to the case of distributed sensors is discussed, based on the theory of statistical hypothesis testing. The development is based on the formulation of a decentralized or team hypothesis testing problem. Theoretical results concerning the form of the optimal decision rule, examples, application to data fusion, and open problems are presented.

A tracking filter for maneuvering sources

It is well known that the extended Kalman filtering methodology works well in situations characterized by a high signal-to-noise ratio, good observability and a valid state trajectory for linearization. This paper considers a problem not characterized by these favorable conditions. A large number of ad hoc modifications are required to prevent divergence, resulting in a rather complex filter. However, performance is quite good as judged by comparison of Monte Carlo simulations with the Cramer-Rao lower bound, and by the filters ability to track maneuvering targets.

Strategies for Distributed Decisionmaking

Several mechanisms which can be used in the real-time distributed coordination of large-scale dynamical systems which display a subsystem structure are derived. A spectrum of approaches using varying degrees of types of communication to reduce the uncertainty of each agent concerning the events controlled by others is presented. The organizational requirements of each coordination strategy in terms of constraints on the interagent relationships are derived, and it is shown that while any of the mechanisms discussed here work well in hierarchical organizational structures, many are adaptable to much larger classes of structures.

Distributed Decisionmaking with Constrained Decisionmakers: A Case Study

A specific distributed decisionmaking problem is considered that reflects essential aspects of tactical command and control teams, particularly communications and limitations to human information processing induced because of time pressure. A constrained optimization problem is fonnulated to maximize overall team performance subject to individual limitations. The results of a systematic investigation of possible solutions are described in terms of performance and workload interactions. In particular, optimal solutions arise where team members act randomly and/or introduce errors. Though results obtained are specific to the case study, a number of more generally applicable principles are evident.

Performance analysis of stochastic timed Petri nets

A performance analysis method is defined and presented for stochastic timed Petri nets (STPNs), a time extension of regular Petri nets that can model systems with concurrency, asynchronous time-dependent protocols, and random processing times. The main result is that the state evolution of a STPN can be viewed in terms of the evolution of another process, the generalized firing process. The performance of the first process can be studied by analyzing the performance of the second. The time-related performance measures for the generalized firing process are obtained by deriving state equations and using these equations to iteratively calculate the steady-state probability distribution. An equivalent discrete-state Markov process is never constructed.<<ETX>>

Modeling Electrocardiograms Using Interacting Markov Chains

A methodology is developed for the statistical modelling of cardiac behaviour and electrocardiograms (ECGs) thai emphasizes (a) the physiological event/detailed waveform hierarchy; and (b) the importance of control and timing in describing the interactions among the several anatomical sub-units of the heart. This methodology has been motivated by a desire to develop improved algorithms for statistical rhythm analysis. Specifically, to develop algorithms that capture cardiac behaviour in a more fundamental way but that stop short of complete accuracy in order to highlight decompositions that can be exploited to simplify statistical inference based on these models. Our models consist of interacting finite-state processes, where a very few of the transition probabilities for each process can take on a small number of different values depending upon the states of neighbouring processes. Each finite-state process is constructed from a very small set of elementary structural elements. We illustrate our methodol...

Performance evaluation of decision-free stochastic timed petri nets

This paper shows how the time-related performance measures of a Decision-Free Stochastic Timed Petri Net can be calculated by recursively calculating the probability distributions of states that describe the system. Sufficient conditions for convergence are given.

Event-based estimation of interacting Markov chains with applications to electrocardiogram analysis

The problem of estimating the state of a distributed finite-state Markov process consisting of several interacting finite-state systems each of whose transition probabilities are influenced by the states of the other processes is examined. The observations on which the estimation procedure is based are continuous signals containing signatures indicative of the occurrence of particular events in the various finite-state systems. The problem of electrocardiogram analysis serves both as the primary motivation for this investigation and as the source of a case study we describe. The principal focus of the paper is on the development of an approach that overcomes the combinatorial explosion of truly optimal estimation algorithms. We accomplish this by constructing a systematic design methodology in which the resulting estimator consists of several interacting estimators, each focusing on a particular sub-process, Important questions that we address concern the way in which these estimators interact and the met...

On the concept of state in decentralized control

An examination of the axiomatic properties of state shows that critical properties of a centralized state do not generalize to the decentralized case. This eliminates a class of possible approaches to decentralized control from consideration, and suggests that any viable approach to decentralized control will require a fundamental reformulation of the concept of state.

Modeling and estimation for a class of multiresolution random fields

Discusses a class of multiresolution models of random fields based on a generalization of the midpoint deflection construction of the 1D Brownian motion. The authors then present least squares (LS) algorithms for the estimation of parameters which define these models and hence provide a framework for synthesizing and analyzing images with fractal-like properties such as those found in statistical representation of natural terrain and other geophysical phenomena. The authors also briefly discuss possible applications of this modeling framework to target detection in images.<<ETX>>