Data Collection for HD Map Updates through MEC-assisted Vehicle Crowdsourcing: A Trade off between Accuracy and Cost

The integration of vehicle crowdsourcing with Multi-access Edge Computing (MEC) has emerged as an efficient way for real-time data collection for high-definition (HD) map updates. In this context, a substantial volume of data must be up-loaded to ensure the precision of HD map updates. However, most existing works ignore differences in vehicle sensing capabilities and upload delays, which pose significant hurdles in achieving both enhanced accuracy and reduced communication costs in HD map updates. This paper addresses the aforementioned issue by offering a comprehensive consideration of vehicle sensing capabilities and upload delays, thereby ensuring the utility of HD map updates within the framework of MEC-assisted vehicle crowdsourcing. We propose an innovative data collection scheme that effectively navigates the balance between accuracy and communication costs while meeting application accuracy and latency requirements. This is achieved by optimizing the vehicle selection process on a small time scale and concurrently adjusting data collection parameters on a larger time scale. Extensive simulation shows that the proposed scheme has superior performance than those baselines.