A Secure Parser Generation Framework for IoT Protocols on Microcontrollers

In the recent years, vulnerabilities found in the packet parsers of Bluetooth Low Energy (BLE) protocol have called for a need to have secure lightweight protocol packet parsers for microcontrollers. Since these packet protocol grammars consist of packets of limited size it is possible to parse them efficiently via Finite State Machines (FSM). However, parsing via FSMs would require developers to either express the grammars via regular expressions or constructed hand-coded parsers. Unfortunately, hand-coding parsers is error-prone; furthermore, due in part to certain constructs found in such grammars which are not commonly found in text-based regular grammars. In addition, expressing binary grammar constructs in regular expression is not only challenging and error-prone but the resulting expressions are often complex and unreadable. Thus the lack of an alternative language for describing these constructs is a hindrance to the use of finite state machines to generate parsers which are safe, secure and computationally bounded. This paper presents a novel secure parser generation framework which consists of an easy-to-use parser description language called "Microparse" and a toolkit that utilizes finite state machines to generate lightweight parsers for micro-controllers. To demonstrate the viability of this approach, we have applied our framework to generate parsers for the BLE protocol running on an Ubertooth One Microcontroller. We demonstrate that the generated FSMs are lightweight enough to be run on devices with very limited resources, and are easier to use for developers; we offer this method as a potential solution for the various bugs found in the implementation of the BLE firmware in the recent years.