Pages

18 August 2017

Army reaps benefits of open-source policy

By: Adam Stone 

The Army’s decision to formalize its open-source software development policy is paying off. At least two major projects have benefited from the policy announced this spring, with open source helping to speed development and save taxpayer dollars, according to officials from the U.S. Army Research Laboratory.

“Open source can reduce development time and lower overall costs, resulting in a win-win situation for the Army and the U.S. taxpayer,” said ARL Deputy Chief Scientist Mary Harper.

When it comes to defense agencies embracing open-source software development, the Army is hardly at the cutting edge. The National Geospatial-Intelligence Agency paved the way with the launch of its GitHub open-source community in 2014. The Navy issued a guidance on the use of open-source code as early as 2007.

Still, ARL leaders say they expect to reap big benefits by formalizing their approach to open source, a term used to describe software for which the original source code is made freely available and may be shared and modified.

“To date, all projects that have been released by ARL as open source have been done under an ad hoc decision-making process. This policy is an experiment by ARL to normalize the process, making open source a day-to-day activity of ARL,” said Cem Karan, an ARL computer engineer who helped formulate the policy.

ARL is publicly disclosing two projects that it says are benefiting from the recently announced policy, known respectively as Dshell and BRL-CAD.

Dshell is an extensible, network forensic-analysis framework and a defensive cyber tool. “In a nutshell, Army networks are under constant assault by various actors. The methods used by these actors change constantly,” Harper said.

“Developing effective countermeasures under short deadlines requires tools that can quickly analyze large amounts of data, and present to analysts only the portions that they need to determine the method of the attack,” Harper continued. “Dshell is a tool that simplifies the defenders’ job.”

Of the 25 contributors on the project, only five are U.S. government or military entities. The remaining outside contributors have been drawn at least in part by the open-source nature of the work, ARL suggests.

BRL-CAD is a modeling system that for more than 20 years has been used to model weapons systems for vulnerability and other analyses. An open-source platform, it is used widely across military, industrial and academic applications.

“Both programs have received contributions from outside of ARL that have significantly accelerated their development,” Harper said. “Without these external contributions, the cost to the U.S. taxpayer would have been increased — to pay for additional personnel — or the work would have been delayed.”

Both projects help to highlight the anticipated benefits of having a more formal approach to open source:

Speed of development. “The open-source development model, which hinges on collaboration with leaders in various fields, has resulted in an unprecedented acceleration of development of ideas and technologies. If ARL doesn’t harness this power, it risks falling behind and becoming irrelevant in the modern world,” Karan said. 
Verification of claims. “Research often depends on the results of custom written software. Research that is published without the accompanying software may not be verifiable or easily repeatable, which can lead to accidentally erroneous claims. By publishing the custom-written software that ARL researchers have developed, outside researchers can verify the quality of ARL’s research,” Karan said. 
Cost sharing. “Research in a particular area may require a tremendous amount of effort developing basic tools before the ‘real’ research can be started,” Karan said. “By publishing these tools as open source, ARL can share the burden of developing the tools with others around the world.” 

Looking ahead, ARL leaders say their explicit embrace of an open-source approach should be a boon to the other military services. But they aren’t yet offering insight into just what those evolutions might look like.

“It’s extremely difficult to guess how the code might be used or leveraged by others,” Harper said. “ARL’s belief is that [the U.S. Army Tank Automotive Research Development and Engineering Center] and other groups will leverage the code as they see fit to support mission and potentially improve ARL’s code at the same time. This has been the case for extremely popular open-source projects such as the Linux kernel, and ARL hopes to leverage this power as well.”

No comments:

Post a Comment