Wurldtech Achilles cyber security testing

When customers can't wait for standards organizations, testing protocols can come from other sources.

By Control Engineering Staff December 2, 2008

What happens when you want to specify product requirements, but there are no relevant standards or testing bodies? Case in point: cyber security. How do you specify and test that your new control system will stand up to attack?

Let’s say you are going to purchase a new control system for some part of your process plant. Being an enlightened buyer, you are concerned with cyber security issues and want to make sure that your supplier has created a robust and secure platform. Your vendor says that of course the platform is robust and secure. So you look for ways to define and quantify those terms in a way that you can both agree. Unfortunately, while much has been written on the topic of what makes a system strong, there is little in the way of approved standards and few options for testing or certification. So what do you do? A group of oil industry companies found themselves in just that situation.

Dr. Nate Kube, chief technology officer of Wurldtech , has specialized in robustness testing and designed approaches to examine a network and its resilience to cyber attack and other bad network conditions. “Before our certification program, there was no real, easy way for the operator to specify a security or robustness requirement, and there was no real way for a vendor to demonstrate that he had a robust device,” he says. “So there was this big gap. There was work being done in standards bodies, ISA SP99 for example, but the completion date was fairly far off. There was no program, immediately available, that the operators could use.”

At the PCSF annual meeting in 2007, Kube made a presentation on testing methods developed from his academic work and industry experience, and how that could be used in a larger testing program. A group of users in critical industries liked what they heard and expressed their support to bring that work together in a form that they could use for testing and product evaluation. Wurldtech began to develop the Achilles program in earnest, and those users started to specify this preliminary testing to their vendors, and the process quickly became formalized.

“We follow IEC and ISO guidelines for proper product certification testing,” Kube adds. “In the absence of a standard, we followed our own experience, some industry pundits, and feedback from large end users to come up with a list of tests that should be executed in order to demonstrate and quantify a controller’s robustness. As the program developed and became more formal, it started to show up in procurement documents. So when a large company wants to put out a bid for a new facility, now Achilles Level 1 testing can be a requirement on the controllers and safety systems.”

In the development of Achilles, Kube worked to move past traditional testing methods, considering them inadequate for the growing sophistication of control applications. As he describes it, “Communication testing in the past has been mostly conformance based, i.e., does your device respond in accordance with some protocol specification? There is a whole realm of negative testing that is being missed, so let’s send a message that’s almost correct, and see how the device responds to that. We’ve found that it’s extremely easy to kill devices, kill their I/O routines, wipe out their programming, and do horrendous things to these devices based on sending slightly malformed network packets. So that’s why we call it robustness and resilience testing. There are increasing instances of converged network topologies where you have the management of safety and control sharing a bus and increased connectivity with the enterprise network. Now you have all these new avenues for malformed or bad traffic, or hacker traffic, to get on to your control network and affect critical systems. So it becomes increasingly important to do negative or robustness testing on the controllers proper. We’re simulating what a hacker or terrorist might take advantage of. Is the protocol implementation robust? Does it handle exceptional conditions? If I send it a bad packet, will the implementation fault? Can I send it high rates of traffic and have it retain its control functionality in such an event? When a device has an Achilles Level 1 certificate, it means that the device is solid at network layers two through four. It will maintain its operational functionality in the presence of some extremely harsh network conditions.”

Based on demand from end users and system builders, Achilles testing has taken off. At present, 12 control platforms have been certified to Level 1, and others are in the works.

ABBPlantguard ControllerAC 800M Controller

EmersonDeltaV Controller

HoneywellExperion PKS C300

ICS TriplexTrusted Controller

IPSTrident ControllerTricon Controller

KongsbergRCU 501 Controller

YokogawaStardom FCJ ControllerProSafe-RS Vnet/IP Safety Control UnitCENTUM CS 3000 R3 Field Control UnitCENTUM CS 3000 R3 Vnet Router

Other organizations are working on similar test strategies, but for the moment, Achilles Certification has established itself as a leading protocol within critical industries.

For more on cyber security, read our new blog with Matt Luallen and Steve Hamburg of Encari .

—Peter Welander, process industries editor, PWelander@cfemedia.com , Process & Advanced Control Monthly eNewsletter Register here and scroll down to select your choice of free eNewsletters .