NASA Spinoff

A Predictive Approach to Eliminating Errors in Software Code

As such data pile up, the heuristics mode can be employed to identify the error-prone code. This problem-solving mode uses an artificially intelligent engine to analyze the metrics and associated error data to learn what metric thresholds predict error within a specific project. (Integrated Software Metrics reports that results are very stable and more precise than those of domain experts who have a strong technical understanding of software analysis.) A user can then continue to use the heuristics mode through most of the software development cycle. Near the end of the cycle, the user can switch to the trend-identification mode to identify trends and find the chronic problems in the system.

The final product, Predictive Server, was released in September 2005. Predictive Server contains all of the features of Lite and Pro, and is scalable for network or distributed deployment. Unlike its predecessors, however, it was developed as a Web-based risk management tool for multiple software projects, and it facilitates collaboration among project managers, developers, and software quality assurance professionals. Essentially, the software aims to meet the demands of having to manage multiple software projects in a networked environment.

When Predictive Server is used on a software project, error data and metrics are compiled and stored in the software’s knowledge database from the very beginning of the project until the end. Thereafter, this historical database of metrics can be used for other software projects in the enterprise. Because Predictive Server is Web-based, all authorized software project teams can access and update the database, helping the organization deliver better software and save development costs.

In late 2005, Integrated Software Metrics announced a 200-percent surge in sales of its Predictive error-prediction tools. Driving this boost were new customers in the telecom, energy, technology, and government markets, including organizations such as Compagnie Financière Alcatel (Alcatel); Chevron Corporation; LogLogic, Inc.; and Northrop Grumman Corporation.

Spinning back to NASA, the Glenn and Goddard field centers are currently using the entire Predictive suite. Both centers rely on the products for critical code that supports NASA’s Earth-orbiting spacecraft.