Publications

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

This report describes the results of a Laboratory-Directed Research and Development project on techniques for pattern discovery in discrete event time series data. In this project, we explored two different aspects of the pattern matching/discovery problem. The first aspect studied was the use of Dynamic Time Warping for pattern matching in continuous data. In essence, DTW is a technique for aligning time series along the time axis to optimize the similarity measure. The second aspect studied was techniques for discovering patterns in discrete event data. We developed a pattern discovery tool based on adaptations of the A-priori and GSP (Generalized Sequential Pattern mining) algorithms. We then used the tool on three different application areas--unattended monitoring system data from a storage magazine, computer network intrusion detection, and analysis of robot training data.

Unattended monitoring systems are being studied as a means of reducing both the cost and intrusiveness of present nuclear safeguards approaches. Such systems present the classic information overload problem to anyone trying to interpret the resulting data not only because of the sheer quantity of data but also because of the problems inherent in trying to correlate information from more than one source. As a consequence, analysis efforts to date have mostly concentrated on checking thresholds or diagnosing failures. Clearly more sophisticated analysis techniques are required to enable automated verification of expected activities level concepts in order to make automated judgments about safety, sensor system integrity, sensor data quality, diversion, and accountancy.

Abstract not provided.

Abstract not provided.

The authors presented a high-level methodology for the design of unattended monitoring systems, focusing on a system to detect diversion of nuclear materials from a storage facility. The methodology is composed of seven, interrelated analyses: Facility Analysis, Vulnerability Analysis, Threat Assessment, Scenario Assessment, Design Analysis, Conceptual Design, and Performance Assessment. The design of the monitoring system is iteratively improved until it meets a set of pre-established performance criteria. The methodology presented here is based on other, well-established system analysis methodologies and hence they believe it can be adapted to other verification or compliance applications. In order to make this approach more generic, however, there needs to be more work on techniques for establishing evaluation criteria and associated performance metrics. They found that defining general-purpose evaluation criteria for verifying compliance with international agreements was a significant undertaking in itself. They finally focused on diversion of nuclear material in order to simplify the problem so that they could work out an overall approach for the design methodology. However, general guidelines for the development of evaluation criteria are critical for a general-purpose methodology. A poor choice in evaluation criteria could result in a monitoring system design that solves the wrong problem.