Modeling In-Network Aggregation in VANETs

The multitude of applications envisioned for vehicular ad-hoc networks (VANETs) requires efficient com- munication and dissemination mechanisms to prevent network congestion. In-network data aggregation promises to reduce bandwidth requirements and enable scalability in large vehicular networks. However, most existing aggregation schemes are tailored to specific applications and types of data. Proper comparative evaluation of different aggregation schemes is difficult. Yet, comparability is essential to properly measure accuracy, performance, and efficiency. We outline a modeling approach for VANET aggregation schemes to achieve objective comparability. Our modeling approach consists of three models, which provide different perspectives on an aggregation scheme. The generalized architecture model facilitates categorization of aggregation schemes. The aggregation information flow model supports analysis of where information is aggregated by a scheme. The aggregation state graph models how knowledge about the road network and its environment is represented by a scheme. Further, it facilitates error estimation with respect to the ground truth. We apply each modeling approach to existing aggregation schemes from the literature and highlight strengths, as well as weaknesses, that can be used as starting point for designing a more generic aggregation scheme.