An edge computing based data detection scheme for traffic light at intersections

Abstract

The reliability and validity of traffic data plays an important role in intelligent transportation systems. Most of the data detection schemes target the network level, and rarely discuss the impact of vehicle data detection at intersections on the intelligent control of traffic lights. Moreover, most data detection is processed in a centralized cloud center, which is not suitable for complex and changeable intersections. In this regard, we propose an edge computing based data detection scheme for traffic light intersections. In our scheme, traffic lights act as edge nodes to detect vehicles data. We first consider a single intersection scenario. We collect vehicle data through V2E communication through base station, and use the quotient filter to detect and verify the validity of the data. We further consider the multiple intersections scenario. We merge the vehicle data of the two adjacent intersections, and then detect the validity of the data by the quotient filter. In addition, the scheme uses mmh3 hash functions in the QF filter to reduce the computing resource occupation of edge nodes and the bit error rate. The experimental results show that the data detection solution is effective in detecting data quickly even there is a large number of vehicles and complex vehicle data. Our proposed scheme can also verify the reliability and effectiveness of vehicles with a smaller delay.