VOIP requires attention to security best practices
- — 17 August, 2007 08:40
New exploits against VOIP continue to emerge, but experts say these demonstrations reveal the need for vigilant security and are not fatal flaws to the technology.
At Black Hat this month researchers released hacking tools against VOIP signaling protocols H.323 and AIX as well as tools to insert audio into VOIP calls. At Defcon, a tool that automatically probes the Session Initiation Protocol for vulnerabilities was released to enable the covert piggy-backing of data over VOIP streams.
The problem lies not in VOIP technology but in its implementation, says Barrie Dempster, a senior security consultant for Next Generation Security Software. "If you apply traditional network security logic to VOIP you can make it as secure as any other protocol," he says.
Much of the notoriety of VOIP vulnerabilities come because the technology is relatively new and its code wasn't necessarily written with security in mind -- a problem that plagues many new technologies.
Dempster cites ways to exploit Asterisk, the open source PBX, including buffer overflows. He says this and other weaknesses can be dealt with by removing the code for unused features and performing security audits on the features that are used. "The problem is not the specific vulnerabilities themselves. It's the maturity of the software. There hasn't been enough security review yet," he says.
The problem is well recognized, and known exploits are publicized to help develop defenses against them. For example, the industry group VOIP Security Alliance publishes a set of hacking tools on its site that it promotes as security tools to test that VOIP gear can withstand real-world attacks.
Securing VOIP is not insurmountable, says Peter Thermos, CTO of security consulting firm Palindrome Technologies. He revealed vulnerabilities to media gateway control protocol (MGCP) that enable rerouting calls or cutting them off. He also showed a vulnerability to ZRTP, a pending-standard, encrypted VOIP protocol that didn't encrypt the sounds of tones made by pressing phone buttons. That potentially left credit card numbers being entered over VOIP lines open to being picked off, he says.
The MGCP problem will ultimately require a change to the protocol itself, but in the meantime users can shore it up by blocking unauthorized access to the ports MGCP uses, Thermos says. The ZRTP problem involved the implementation of the protocol and has been addressed with a patch.
The best route for businesses implementing VOIP is to set individual security requirements ahead of time, which differ among companies, he says. A financial institution or government agency may need confidentiality and therefore more encryption than other businesses, he says.
"The common mistake I see is that customers don't define their security requirements for their particular network, realize later that they need security, then perceive it as an additional cost," Thermos says. Getting security tools in place from the outset also better defends VOIP against threats not yet discovered, he says.