Abstract
<title>Abstract</title> <p>This paper proposes an enhanced version of the AODV routing protocol forVehicular Ad Hoc Networks (VANET), based on the integration of advanced link quality metrics and an intelligent classification model called LSENN (LinkState Estimation Neural Network). Unlike conventional AODV, which relies mainly on hop count, the proposed approach incorporates physical-layer parameters such as inter-vehicle distance, vehicle speed, RSSI (Received SignalStrength Indicator), and OFDM (Orthogonal Frequency Division Multiplexing) decoding error to better assess link reliability. The LSENN model classifies links into three states: connected, transitional, and disconnected, enabling more informed and reliable routing decisions. Simulation results demonstratet hat AODV-LSENN significantly improves packet delivery ratio, reduces end-to-end delay, and lowers jitter, particularly in low to medium density network scenarios. However, performance gains become limited in high density scenarios due to increased channel congestion and MAC-layer collisions.</p>