Research

 Knowledge management for individual and organizational performance

• Capturing and organizing tacit knowledge and applying it in context to solve future problems
• Case based reasoning as a knowledge management tool for organizations
• Performance support in cross-unit maintenance activities
• Domain-specific problem solving environments
• Using ambient intelligence design ideas to create physical spaces that assist in diagnostic and troubleshooting tasks and learn from actions performed in them

Organizational dynamics

• Modelling knowledge assets as organizational memory
• Computational modelling of organizational dynamics
• Knowledge management techniques for virtual organizations that facilitate problem solving using both signal information and tacit knowledge.
• Role of organizational memory in innovation and problem solving
• Knowledge management aspects of goal and priority coordination between cognate agencies
• Interfaces to knowledge management systems that are integrated into work environments such as tangible interfaces.

Smarter instruments and sensor networks

• Associating semantic information with instruments and sensors to characterize the sources of observed data, and adding semantic content to digital signals as they are acquired
• Connecting semantically enriched data streams to analytical and archival components in computing and storage Grids
• Sensor networks for ecological monito ring of freshwater lakes and coral reefs
• Distributed agent-based signal processing to embed condition monitoring and intelligent response capability in the network
• Smart devices and machine reasoning about device characteristics
• Real-time sensor networks operating over segmented networks with different types of media and different performance characteristics

Some research projects

• On-line instruments and sensor networks
  • Common Instrument Middleware Architecture
  • Automating Scaling and Data Processing in a Network of Sensors: Towards a Global Network for Lake Metabolism Research and Education
  • Agent-Based Systems for Distributed Signal Processing and Condition Monitoring
  • Web Portals as User Interfaces to Networked Instruments and Sensors
  • EmbedID: Contextualizing Data from Sensors and Instruments
• Knowledge Management for Performance Support in Distributed Organizations
  • Knowledge Management for Fleet Maintenance: The Knowledge Projection System
  • DesignWorks: Supporting Collaborative Problem Based Learning in Distance Education
  • Research on Interagency Coordination in International Science and Engineering Research: Implications for Policy Formation & Scientific Collabora tion
  • Diagnostic Workbench

On-line instruments and sensor networks

NSF Middleware Initiative – Common Instrument Middleware Architecture

Scientific instruments and sensors are crucial to scientific advancement.  They provide the raw observations used to develop, verify, and falsify theories. Data from instruments typically have an extensive lifecycle, which includes reduction, annotation, analysis, and storage. Standards exist to facilitate this process everywhere except at the head of the pipeline, at the instrument where data is acquired. The Common Instrument Middleware Architecture (CIMA) project, supported by the National Science Foundation Middleware Initiative, aims at “Grid enabling” instruments as real-time data sources to improve accessibility of instruments and to facilitate their integration into the Grid.  A primary challenge addressed by this research program is the lack of a generalized approach to instrument middleware that allows existing and new instruments to be integrated into Grid computing environments. Other issues being explored include extending the accessibility of instruments to new classes of users, use of instruments by software agents, and increasing the longevity, flexibility and durability of software systems for instruments.

Automating Scaling and Data Processing in a Network of Sensors: Towards a Global Network for Lake Metabolism Research and Education

Key problems in fielding ecological sensing networks are scalability of the network is terms of adding new stations and sensors, and in automatic detection of events in real time from data flowing from sensor stations. This project, joint with researchers at the University of Wisconsin, the University of California San Diego, and SUNY Binghamton, is developing new methods for automatically updating data flows from sensors to biological databases to improve the scalaibility and manageability of large ecological sensor networks. We are also looking at novel ways to detect and respond to events (specific patterns of sensor readings). The laboratory for this research is a network of freshwater lake observatories in the U.S. and in the Asia-Pacific region.

Agent-Based Systems for Distributed Signal Processing and Condition Monitoring

We are developing a software system, Sensor AGEnt Execution Environment (SAG3E), that allows users to easily implement production quality event detection and signal processing agents in sensor networks. These networks provide streams of raw and processed sensor data on wide area networks. New nodes that process or respond to data flows can be added and managed by groups operating in concert or independently. Development and deployment of agents is facilitated through the use of design templates that assist users in acquiring and manipulating existing data streams, and in creating new streams or value-added services based on them. Potential applications include environmental or process sensing, condition monitoring and alerting, building system monitoring, condition-based maintenance, and other applications where multiple, possibly independent groups or organizations may need to acquire and react to data from real-t ime sources.

Web Portals as User Interfaces to Networked Instruments and Sensors

This project is a collaboration with the NSF Middleware Initiative’s Open Grid Computing Environments (OGCE) group to develop GridSphere portal server plug-ins to support remote access to instruments and sensors. The primary deliverables are GridSphere-compliant portlets for remote instrument access and data management.

EmbedID

A key problem in contextualizing data from sensors and instruments is the need to provide metadata at the source of the data. This represents an ontological problem of describing in a well grounded manner the signal semantics, source characteristics, and environmental attributes relevant to the “raw” data. The EmbedID project aims at developing ontologies or structured vocabularies to describe sensor and signal attributes in a manner that will allow external software and hardware to interact with the device, contextualize the data coming from it and reason about transforming this data. A particular feature of these vocabularies is that they can be used in machine reasoning tasks, such as semantic mediation and agent-like device-device interactions. Potential applications include device aware location independence and automatic fusion of sensor data streams.

Knowledge Management for Performance Support in Distributed Organizations

Knowledge Management for Fleet Maintenance: The Knowledge Projection System

A critical problem in distributed problem solving within an organization (and across organizational boundaries in the case of virtual organizations) is that knowledge is in many forms and in many locations. A particularly useful form of knowledge, personal experience, is typically difficult to focus and its use is highly dependent on the structure and function of social networks within an organization. The Knowledge Projection System approach to leveraging tacit knowledge is to “observe” problem-solving activities engaged in by experts and automatically capture details of each problem solving session in a case base. When applied to a new problem, the case base can be queried and related cases used to drive diagnosis and provide or suggest a solution to the current problem.