IoT-D: Towards Internets of Dialect-Speaking Things
Abstract
IoT-D: Towards Internets of Dialect-Speaking ThingsSummaryMotivation In Naval environments, an increasing number and variety of IoT devices, such as sensors, microcontrollers, electronic appliances, and computers are interconnected, collecting, processing, and sharing dataamong them and assisting in the operations of Navy personnel and physical systems. Like machines on the Internet, interconnected IoT devices are subject to cyber attacks. Of particularly concern are attacks that infiltrate, eavesdrop, disrupt, or hijack communications between IoT devices, most notably man-in-the-middle (MITM) attacks. The wireless (e.g., Bluetooth) or broadcast (e.g., CAN Bus) nature of IoT communications makes such attacks easy to launch. A main culprit behind these attacks is the standard protocols which enableinter-networking among heterogeneous devices. Such protocols raise two main concerns: (1) Their standard specifications are universally adopted, making it easy for attackers to learn, understand, and forge protocol messages; (2) Their generality leads to an implementation bloat, enlarging the attack surface and increasing the number of vulnerabilities in them.Proposed Research To address the above concerns, we are motivated to break the protocols??? uniformity and bloat, by creating Protocol Dialects, each being a usage-specific, debloated (in both ???depth??? and ???breadth???) variant of the standard IoT protocol stack, with mutated protocol message formats and message-exchange state machines. We call such a paradigm Internets of Dialect-Speaking Things or IoT-D, with the following advantages: (1) Compartmentation: devices in each IoT-D ???speak??? a dialect that can only be understood amongthemselves, achieving controllable isolation; (2) Efficiency: a debloated protocol stack achieves higher time and space efficiency; (3) Security and Agility: MITM attackers will not be able to understand and forge dialect messages and, thanks to on-the-fly dialect switching, they will not have sufficient time to learn a dialect; (4) Transparency: a dialect is transparent to applications above and hardware below and can inter-operate with the standard protocol via a translator.In this project, we propose to develop an IoT-D-enabling framework, covering four major steps in protocol dialect generation: (1) protocol stack analysis and modeling; (2) protocol stack debloating and flattening; (3) protocol stack mutation; and (4) dialect translation. Our framework is based on a formal, elegant model called ProtBNF, which uniformly models multiple aspects of a protocol stack. Architecturally, the proposed framework will be concretize as an IoT-D Dialect Factory. The input of the factory is the specification andimplementation of an original protocol stack and the output is a series of dialects (and corresponding translators) for a target IoT-D. To demonstrate our framework???s practicality, we will perform case studies on at least fourIoT protocol stacks, including Bluetooth and CAN Bus.Innovation Claim and Impacts To the best of our knowledge this project is among the first to create protocol dialects for specialized IoT (and embedded) environments, with a uniform, well-defined model guiding protocolanalysis, debloating, mutation, and translation. The proposed IoT-D framework reflects new synergies among the areas of program analysis/transformation, protocol engineering, and moving target defense. If successful,this project will lead to significant impacts. Technically, it challenges and changes the ???uniformity and bloat??? status quo of current network protocol development and deployment. Practically, it will help elevate thelevel of security and resiliency of specialized IoT and embedded networks that are widely deployed in Naval environments. Our ???academia+industry??? project team is strongly committed to transferring technologies developed in this project for Naval field assessment and potential deployment.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 27, 2018
- Source ID
- N000141812674
Entities
People
- Dongyan Xu
Organizations
- Office of Naval Research
- United States Navy
- University of Virginia