Robust Safety for Autonomous Vehicles through Reconfigurable Networking

TL;DR

This paper proposes using Software Defined Networking (SDN) to enhance the robustness and resiliency of in-vehicle networks in autonomous vehicles, ensuring safety functions remain operational during failures.

Contribution

It introduces a Software Defined In-Vehicle Networking (SDIVN) architecture that improves failure resilience without adding overhead, validated through simulations and testing.

Findings

SDIVN matches legacy networks in performance under normal conditions.

SDIVN outperforms legacy networks during link failures.

The architecture ensures timely message delivery during failures.

Abstract

Autonomous vehicles bring the promise of enhancing the consumer experience in terms of comfort and convenience and, in particular, the safety of the autonomous vehicle. Safety functions in autonomous vehicles such as Automatic Emergency Braking and Lane Centering Assist rely on computation, information sharing, and the timely actuation of the safety functions. One opportunity to achieve robust autonomous vehicle safety is by enhancing the robustness of in-vehicle networking architectures that support built-in resiliency mechanisms. Software Defined Networking (SDN) is an advanced networking paradigm that allows fine-grained manipulation of routing tables and routing engines and the implementation of complex features such as failover, which is a mechanism of protecting in-vehicle networks from failure, and in which a standby link automatically takes over once the main link fails. In this paper, we leverage SDN network programmability features to enable resiliency in the autonomous vehicle realm. We demonstrate that a Software Defined In-Vehicle Networking (SDIVN) does not add overhead compared to Legacy In-Vehicle Networks (LIVNs) under non-failure conditions and we highlight its superiority in the case of a link failure and its timely delivery of messages. We verify the proposed architectures benefits using a simulation environment that we have developed and we validate our design choices through testing and simulations