RESILIENT AND SECURE SERVICES IN INTERNET OF VEHICLES

Abstract

Internet of Vehicles (IoV) integrates a series of network services to help vehicles conduct necessary tasks related to driving and vehicle automation. In this thesis, we propose resilient and secure service placement and service management in IoV. We first address the dynamic service placement problem using deep reinforcement learning (DRL) to provide better services by relocating them to optimal locations when the network state change. We also address the problem of adversarial poisoning attacks and propose an anomaly detection framework that learns and detects illegitimate behavior without having a-priori training dataset. We also address the problem of service outages due to attacks on the edge network and to ensure disruption-free service availability, we propose an attack-resilient optimal service placement and ensure load-balancing to enhance the network performance when the outage attack is active, and recovery is not completed. We then investigate an online service lifecycle management problem in this thesis that works in conjunction with varying service demands and performs service lifecycle decisions. This thesis carries out a comprehensive simulation study on our proposed solutions to demonstrate their effectiveness.