Improving the quality of service by continuous traffic monitoring using reinforcement learning model in VANET

Abstract

The growing number of automobiles on the road has now become a significant source of traffic, accidents, and pollution. Intelligent Transportation Systems (ITSs) could be the key to finding solutions that drastically reduce these issues. The linked vehicular networks channel is a fast-expanding topic for workflow management system and development. Traffic detection is a big issue on city streets. To make informed decisions in order to prevent traffic jams, one of the solutions is the Vehicular Ad-Hoc Network (VANET). We describe an approach for detecting traffic jams in both vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication, which provides vehicle drivers with multiple alternatives for determining the location of congestion, its size, and how to prevent becoming stuck in a traffic jam. The data are sent to the driver to assist him in making the appropriate selection. Delivering Quality of Service (QoS) for automotive networks is a difficult challenge due to the characteristics such as terms of transmission or high mobility, congested and fragmented channels, hardware defects, and a large number of vehicular devices. As a result, it is extremely desired to get and distribute resources efficiently. This research uses Reinforcement Learning with Decision-Making Model (RLDMM) method to enhance channel allocation. By using latency, Signal-to-Interference Ratio (SIR) and QoS, the available channel is initially determined. This method is used to determine suitable channel for platoon members. As a result, the proposed RLDMM achieves an SINR reduction of 29 Db, 69 kbps of throughput, 21.4% of collision probability with 4[Formula: see text]ms of latency.