Deterministic Task Scheduling in In-Vehicle Networks for Software-Defined Vehicles

TL;DR

This paper proposes a deterministic task scheduling method for in-vehicle networks in software-defined vehicles, improving reliability, workload balance, and supporting increased computational demands.

Contribution

It introduces a novel deterministic scheduling approach that outperforms traditional methods in guaranteeing service levels and workload distribution in IVNs.

Findings

Deterministic scheduling better guarantees service levels than shortest path or minimal execution time methods.

Supports increasing computational workloads with balanced resource utilization.

Effective across various IVN topologies and hybrid wireless-wired implementations.

Abstract

Modern vehicles are embedding increasing levels of automation, connectivity, and intelligence, which require advanced in-vehicle networks and computational platforms to support the dependability and deterministic requirements of critical in-vehicle functions. To this end, the automotive industry is shifting towards software-defined vehicles (SDVs) and zonal E/E architectures with centralized computing nodes. Realizing the full potential of these new architectures requires an efficient management of the in-vehicles computational workload. In this context, this paper introduces a deterministic task scheduling approach for in-vehicle networks (IVN), and demonstrates that it can better guarantee deterministic service levels than alternative approaches based on the shortest path or the objective to minimize task execution time. Our evaluation also demonstrates that a deterministic task scheduling can satisfactorily support increasing in-vehicle computational workloads and tasks, and achieve a more balanced workload and resource utilization across the IVN. These gains are validated across a variety of IVN topologies, and in hybrid wireless-wired IVN implementations, where a gradual introduction of wireless offers increased in-vehicle connectivity diversity.