Designing a Functional Programming Architecture for the Internet of Things

Abstract: As the Internet of Things (IoT) grows, so too do security concerns: as well as typically having access to sensors and actuators, IoT devices are often programmed using bug-prone, low-level languages. Such a combination results in vulnerabilities that pose risks to privacy and safety. This thesis aims to address this by making it possible to run high-level functional programs on IoT devices, a daunting prospect with traditional hardware due to the overheads of functional programming runtimes. To accomplish this, an architecture and partial implementation of a "natively functional" processor for IoT, named Cephalopode, is presented. The processor performs both graph reduction and garbage collection directly, without requiring an expensive software runtime. Implementing Cephalopode raised several opportunities for improving the process of hardware design. To that end, this thesis presents the finite state machine editor Stately and the high-level language Bifröst. Stately raises the level of abstraction of finite state machines enough to avoid a proliferation of edges during design, while maintaining efficiency and low-level control. Bifröst offers a higher-level approach to hardware design, allowing complex algorithmic processes—in particular those that communicate extensively with other components—to be described in an imperative language and compiled to an RTL-level circuit model.