DARPA prescribes funds for self-healing systems R&D

DARPA prescribes funds for self-healing systems R&D


The Defense Advanced Research Projects Agency will fund up to three $5 million trials of software probes and gauges for large systems by fiscal 2003.

DARPA wants to see software components 'fit within tolerances,' which they cannot now do, John Salasin of the agency's Information Technology Office said this month. 'We're looking for a semiformal means to monitor large-system performance.'

Gain from failure

In peer-to-peer architectures of the future, he said, one software component's failure could be compensated for by other software, in contrast with client-server components that often deadlock and crash. The system would theoretically become self-healing.

Speak the language of Dynamic Assembly for Systems Adaptability, Dependability and Assurance

' ADL: Architecture Description Language

' AIDE: Active Interface Development Environment

' ArchDiff: Architecture differencing tool that calculates the differences between two xADL representations

' FleXML: Flexible XML, an extension of XML

' Gaugent: Mobile agent that serves as a system gauge

' UML: Unified Modeling Language

' Workflake: Java applet that integrates worklets into a workflow

' Worklet: Mobile agent that makes configuration changes to a range of host information systems

' WVM: Worklet Virtual Machine that attaches to target components and exchanges worklets with its peer WVMs

' xADL: ADL based on XML

' xArch: XML-based representation for software architectures

' XML: Extensible Markup Language

' XUES: XML-based Universal Event Service

DARPA expects its Dynamic Assembly for Systems Adaptability, Dependability and Assurance (DASADA) effort to produce the probes and gauges, which are cross-platform Java applets, plus an adaptation engine that can dynamically reassemble software components without taking a system down.

The probes would collect data while a system is running under frameworks such as Enterprise JavaBeans, Java2 Enterprise Edition, Common Object Request Broker Architecture, and Microsoft's Component Object Model, Distributed COM and .Net.

Fundamental differences between these frameworks are a leading cause of system incompatibilities, Salasin said.

The probes and gauges would measure whether specific components continue to fit within functional tolerances as a system runs and as it undergoes change. Gauges might appear on screen as small graphs, clocks, progress bars or traffic lights.

Nineteen companies and universities are working on DASADA tools, which Salasin said must be built on commercially available software frameworks.

The DASADA tools also must be able to measure the actions of system connectors as well as a system's own actions.

Seven Defense Department and intelligence activities are evaluating the early DASADA tools for real-world problems such as integrating a new weapons system into existing aircraft software, which Salasin said now takes up to 12 months.

Two goals in sight

One target for DASADA efforts is the Airborne Warning and Control System used in E-3 AWACS surveillance aircraft.

Another target for DASADA efforts is the Navy Pacific Command's GeoWorlds, a loosely coupled intelligence collection system.

An index of current DASADA tools appears at www.rl.af.mil/tech/programs/dasada/tools.html.

inside gcn

  • artificial intelligence (ktsdesign/Shutterstock.com)

    Machine learning with limited data

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group