Multiple Vehicles Collaborative Data Download Protocol via Network Coding Data services (i.e., office on wheels and entertainment on wheels) are expected to become a primary driver in the development of future connected cars. However, the sparse spatial distribution of roadside stationary units (RSUs) along the road renders the downloading of data via roadside-to-vehicle (R2V) connections intermittent. As a result, data services, particularly for those dealing with large volumes of data, may not achieve a good quality of service. In this paper, we propose a multiple-vehicle protocol for collaborative data downloading by using network coding (NC). When multiple vehicles that are approaching each other have a common interest in certain data, they can collaboratively download the data from an RSU to significantly reduce their download time. We first derive the probability mass functions (PMFs) of the downloading completion time for three downloading methods, i.e., random, feedback, and NC based, to quantify the benefits of the proposed scheme. Our analytical derivations show that, compared with random- and feedback-based methods, the proposed approach can significantly improve the download time and will remove any need for uplink communications from vehicles to the infrastructure. Moreover, we discuss the cooperative group formation issues and vehicle-to-vehicle (V2V) data sharing in detail. Simulation results show that the proposed protocol has a more robust performance compared with random- and feedback-based schemes. In addition, we constitute simulations to show that the proposed scheme can apply to the scenarios with dynamic network topology and imperfect V2V data sharing.