Vehicle to Infrastructure(V2I) communications aim to provide mobile users on the road low-cost Internet and driver safety services.However, to meet Quality of Service(QoS) requirements of various applications and efficiently utilize limited wireless channel resourc-es, the transport layer protocol has to perform effective rate control in low channel quality and frequent changing topology communica-tion environment. In this paper, we propose a novel rate-control scheme in infrastructure based vehicular networks that avoids conges-tion and starvation and promotes fairness in end-to-end V2I communications. In vehicular networks, a bottleneck roadside unit(RSU)keeps track of its buffer size, aggregate incoming rate, and link throughput, and appropriately allocates bandwidth to traversing flows.With feedback information from the RSU, source nodes dynamically adjust their sending rates to avoid buffer overflow or starvation atthe bottleneck RSU. Simulation results show that the proposed scheme can reduce not only packet losses owing to buffer overflow butalso buffer starvation time, which improves the utilization efficiency of wireless channel resource. Vehicle to Infrastructure (V2I) communications aim to provide mobile users on the road low-cost Internet and driver safety services. Now, to meet Quality of Service (QoS) requirements of various applications and efficiently utilize limited wireless channel resourc-es, the transport layer protocol has to perform effective rate control in low channel quality and ever changing topology communica- tion environment. In this paper, we propose a novel rate-control scheme in infrastructure based vehicular networks that avoids con-tion and starvation and promotes fairness in end -to-end V2I communications. In vehicular networks, a bottleneck roadside unit (RSU) keeps track of its buffer size, aggregate incoming rate, and link throughput, and appropriately allocates bandwidth to traversing flows.With feedback information from the RSU, source nodes dynamically adjust their sending rates to avoid buffer overflow or starvation atthe bottleneck RSU. Simulation results show that the proposed sche which can reduce not only packet losses due to buffer overflow butalso buffer starvation time, which improves the utilization efficiency of wireless channel resource.