Distributed Upstream Data Cleaning in VANET

Abstract

The mobile road stations in Vehicular Ad~hoc NETworks (VANET) are generating huge quantities of traffic-related data over wireless communication medium. For different business purposes, data has to be collected by heterogeneous distributed Road Side Units (RSUs) and pushed to traffic agencies (i.e., back-end) to be processed. Commonly, servers run a heavy processing in schema and data instance levels to ensure data transformation and data cleaning prior to storage in a production data warehouse. In particular, a comprehensive treatment is dedicated to detect and remove data redundancy and filter out or correct outlier instances. In order to alleviate these challenging tasks, a new proposal to delegate part of this data processing to the network edge in the RSUs is going to be investigated. Indeed, the RSUs are subject to potentially capture redundant data as well as defective measurement values from field data sources. Through a cooperative approach, RSUs could perform a near-real-time upstream filtering and cleaning treatment of the gathered data. Hence, an overview to tackle the upstream data cleaning and redundancy removal in a distributed fashion is presented. The proposed approach is expected to reduce the complexity of the data cleaning task and to scale better with the network size and its resulting data.