Hybrid Vehicular and Cloud Distributed Computing: A Case for Cooperative Perception

TL;DR

This paper introduces a hybrid offloading framework for vehicular and cloud computing that optimizes resource use and improves processing rates for cooperative perception tasks through combined LTE and V2V communication.

Contribution

It presents an optimized resource assignment and scheduling framework for hybrid offloading in vehicular networks, enhancing processing efficiency for cooperative perception.

Findings

Significant increase in processing rate with hybrid offloading.

Processing rate improves with increased V2V connectivity.

Framework effectively utilizes edge and micro cloud resources.

Abstract

In this work, we propose the use of hybrid offloading of computing tasks simultaneously to edge servers (vertical offloading) via LTE communication and to nearby cars (horizontal offloading) via V2V communication, in order to increase the rate at which tasks are processed compared to local processing. Our main contribution is an optimized resource assignment and scheduling framework for hybrid offloading of computing tasks. The framework optimally utilizes the computational resources in the edge and in the micro cloud, while taking into account communication constraints and task requirements. While cooperative perception is the primary use case of our framework, the framework is applicable to other cooperative vehicular applications with high computing demand and significant transmission overhead. The framework is tested in a simulated environment built on top of car traces and communication rates exported from the Veins vehicular networking simulator. We observe a significant increase in the processing rate of cooperative perception sensor frames when hybrid offloading with optimized resource assignment is adopted. Furthermore, the processing rate increases with V2V connectivity as more computing tasks can be offloaded horizontally.