Panel: Internet of Things (IoT) Security

Moderator: David Balenson, SRI International

Abstract:
In the Internet of Things, any device may be connected to the Internet.  In this panel, the speakers advocate different approaches for securing IoT devices.  Do traditional security models still apply?  Is it feasible to address security at the scales that IoT requires?  Does IoT require new architectures?  New types of security policies?  New ways of looking at device and platforms?

Panelists:

Dr. Michael Clifford, Principal Security and Data Scientist, Noblis NSP  SLIDES
For end users, the Internet of Things holds the potential for the coordination of devices, sensors, networks, systems, and all of their resulting data.  But from a security perspective, Internet of Things brings the challenges of billions of low-cost, unpatched, and unmaintained devices.  For sufficiently inexpensive devices, manufacturers lack the financial incentives to minimize device security risks over time.  As the boundaries between devices used for different roles within people’s lives dissolve, traditional defense in depth strategies may no longer provide acceptable levels of risk.  When attackers can move laterally between unprotected devices with ease, a new approach to security may be needed. I will discuss the risks posed by vulnerabilities in common devices, as well as tradeoffs in traditional security approaches when applied to IoT devices.  These tradeoffs can motivate approaches such as the use of observation-based trust models and reputation systems or the abstraction of devices into device classes, against which dynamic, contextual security policies can be applied.
 
Dr. Michael Collins, Chief Scientist, RedJack  SLIDES
Back when we called IoT devices embedded systems, we were dealing with common mode flaws -- vulnerabilities in Rompagerwebservers or Samsung televisions have been around for a good while.  The real problem is that the security flaws exploited, such asthe Mirai botnet's default password exploit, are not new to security people - just new to a fresh population of developers and their devices. A network administrator cannot wait for IoT devices to be secure, cannot assume IoT devices will be designed securely, and cannot rely on patching to secure IoT devices.  Rather, we have to assume new forms of defense, predicated on isolating and compensating for IoT failures.  I will discuss several approaches towards maximally locking down devices in a working network, compensating for common IoT failures, and the implications for long term network security.  
 
Dr. Ulf Lindqvist, Program Director, SRI International
IoT devices span a wide spectrum of sophistication and computing resources and are being deployed in unprecedented numbers. These devices are often trusted with our critical tasks and information, regardless of how trustworthy those devices actually are or how poorly they are managed. We need new methods to automatically and continuously assess the level of security of IoT systems and networks, because if we cannot tell how changes to configurations or devices impact overall security, how will we be able to improve security with any degree of confidence? Also, we need such assessment methods so that we can design and select the most effective security mechanisms for large numbers of heterogeneous IoT devices.