A Proposed Software Protection Mechanism for Autonomous Vehicular Cloud Computing

Abstract

Cars are becoming smarter every day. They are being equipped with sensors, communications interfaces, and more powerful processing capabilities. The primary purpose was to enable these cars to drive themselves without any human interventions and become so-called Autonomous Vehicles (AVs). But why stop there, why not harness all that computing power for a greater collective purpose. That’s how the idea of Autonomous Vehicular Cloud Computing (AVCC) was born. Nonetheless, this is not a trivial task, the mobile and dynamic nature of vehicles poses a significant challenge in the formation and management of this cloud computing model and yet a more substantial challenge in terms of security and privacy of all the parties involved in this system. In this paper, we focus on protecting software running on AVCC. We use dynamic obfuscated compilation to complicate programs’ execution paths and hinder information leakage via side channels attacks. Relying on compilers offers advantages, such as the independence of architecture and support for a variety high-level programming languages and application simplicity with minimal set-up cost. Here, we introduce our system in the realm of ARM processor, which power AVCC. Then, we present execution statistics for simple standard programs. The results show tangible timing variations in diversified code versions for the same program, which may disrupt side-channel attacks.