Navy looks to harden legacy software

A Government Accounting Office report released in May 2016 found that federal agencies rely far too heavily on legacy computing systems and software for critical missions, including a Defense Department's nuclear command and control system that was still using floppy disks from the 1970s.

Replacing such systems is a top priority governmentwide, but the Office of Naval Research is also exploring a different approach.

An ONR research grant will seek to secure outdated military software and hardware. University of California-Riverside researchers said they would look for ways to "harden" code that is no longer maintained but remains in daily use by military agencies.

The GAO report noted that operators of the Strategic Automated Command and Control System were expected to update storage and add processing power by the end of fiscal 2017. The system used to coordinate U.S, nuclear forces still relies on legacy software written for 1970s-era IBM Series/1 mainframes.

Hence, the UC-Riverside investigators will attempt to secure decades-old code as part of a five-year, $4.68 million research grant. The primary focus will be pushing the boundaries of "binary analysis," said Heng Yin, an associate professor of computer science and engineering.

Yin and co-investigator Chengyu Song will initially focus on widely used software programs such as Adobe Acrobat Reader and Microsoft Office to develop the prototype and a method for securing similar legacy systems. "We want to identify program features and replace things in the binary code, then secure the [executable] binary," Yin explained in an interview.

Widely used applications such as Acrobat Reader have become a magnet for hackers. Adobe regularly releases patches for the document reader along with its Flash player and other utilities. Many recent patches have focused on exploits that take advantage of code execution vulnerabilities. The ONR initiative seeks to develop new security approaches beyond patches as a way to harden code, the researchers said.

One common exploit is embedding malicious code in a PDF document or other attachment. Once clicked, the exploit spreads rapidly across networks. Yin said one hardening approach involves the widely use JavaScript code underlying applications like Acrobat Reader. In this example, the JavaScript runtime would be removed as a way of hardening the executable code while preserving the document viewing application and related functionality.

Another vulnerability associated with legacy software is corruption of computer memory. Securing the executed code could also help shield software and hardware components, Yin said.

Once hardening techniques have been demonstrated, the challenge will be scaling those methods to address a host of common DoD applications. Part of the research effort involves identifying metrics and then scaling possible fixes. "This is an open problem," Lin said. "We provide our ideas for solving problems" with the potential to be scaled.

Lin estimates that the security effort will be about 80 percent research and 20 percent development. The university researchers have a firm three-year commitment from ONR with two option years. After that, Lin said, industry must step up to commercialize the hardening approach with solutions that are "stable, reliable and scalable."

A Navy spokesman declined to comment on the research effort.

An ONR solicitation released last year noted "the size and complexity of software in [Navy] systems has grown tremendously in recent years," resulting in what program officials called "bloat." Left unchecked, that complexity "often leads to an increase in unexpected software execution failures and vulnerable attack surface for software exploitation."

Program officials said they are particularly concerned about commercial software that is increasingly used in Navy systems.

Adding to the complexity of COTS software is the trend toward code reuse, an approach aimed primarily at boosting programmer productivity. The downside is additional layers of software libraries and APIs that increase "complexity and makes efforts to [determine] the safety or security of that software even more intractable," the request for proposals warns.

Hence, the university researchers will focus on reducing software "bloat" in legacy Navy systems to improve security without sacrificing functionality.