Dynamic Hierarchical Aggregation for Vehicular Sensing

Abstract

Vehicular sensing has gained prominence in recent years with its use in entities, including traffic management centers, forensic authorities, and air pollution control units. It also provides end users with real-time street images, parking summaries, and road congestion status. To reduce bandwidth usage and improve the content value, the sensed data must be aggregated. Data aggregation is said to be efficient when the destination (i.e., a node that serves as a data collection point in the network) is capable of receiving sensed data from a significant proportion of vehicles. However, when a large number of vehicles attempt to send sensed data, the network becomes congested eventually causing packet losses and collisions. Thus, if aggregation is performed without considering key factors, such as number of vehicles and network dynamics, it is difficult to ensure the efficient collection of sensed data at the destination. In this paper, we propose a dynamic hierarchical aggregation scheme in which sensed data is aggregated using a hierarchy. Moreover, the hierarchy is dynamically updated based on theoretically estimated delivery efficiency. In particular, we perform partition and merge operations within the hierarchy to achieve an improved value of delivery efficiency. The simulation results show that the proposed scheme ensures efficient data collection even with stringent delay requirements and achieves scalability with respect to a number of vehicles in the network.