📝 SummarXiv

Abstract

In multi-core real-time vehicle control systems, synchronization blocking and resource contention pose critical challenges due to increasing task parallelism and shared resource access. These issues significantly degrade system schedulability and real-time performance, as traditional task allocation algorithms often overlook blocking impacts, leading to high scheduling failure rates under heavy loads. To address this, we propose the BR-WFD algorithm, which integrates blocking time estimation and resource similarity analysis. The algorithm minimizes global blocking overhead by prioritizing tasks with high synchronization sensitivity and aggregating shared-resource-accessing tasks onto the same core. Extensive simulations show that BR-WFD reduces required processor cores by 11\% to 28\% and maintains a 15\% to 20\% higher schedulable ratio compared to traditional methods under high-load and resource-competitive scenarios. This demonstrates its effectiveness in enhancing real-time performance and resource efficiency for multi-core task scheduling in intelligent driving systems.