In this study, we investigated an algorithm that combines GPS and Doppler migration technologies to improve the collision prevention technology for judging vehicles’ directions on a highway interchange. In this algorithm, vehicle-to-vehicle communication was accomplished using transceivers installed on the vehicles, and the investigated algorithm analyzed the characteristics of Doppler shift values of the frequency offset sent by the moving vehicles to observe the dynamic relationships between an observer vehicle and other vehicles. From the change in frequency offset (Doppler shift), the investigated system can detect the situation of multiple merging vehicles from a highway interchange and analyze whether these vehicles can safely enter the main lane. In this work, we studied and discussed the investigated collision avoidance warning system by using vehicle-to-vehicle communication in non-line-of-sight situations via the 5G mobile network. The simulations and analyses were mainly carried out under the situations described above to observe the changes in Doppler shift values. In this research, a driving collision theory is proposed for judging whether a collision will occur by the driving collision warning judgment process, and the investigated collision warning mechanism is verified.