Veins 5.0 and INET 4 are using completely separate models for simulating wireless channels. While it is perfectly feasible to make both models aware of "correct" vehicle positions (by instantiating both model stacks, that is, using two applications, two wireless card modules and two mobility modules per vehicle -- as well as two channel modules per simulation, though only one connection to SUMO), this is not necessarily the easiest way forward.
Are you sure your study needs to simulate both types of models in the same simulation? If not, build two separate simulations -- one using Veins+veins_inet+INET (using only the latter's wireless models), one using Veins.
Are you sure you cannot rely on an INET WLAN model configured to opMode=p
instead of the wireless models of Veins? See https://github.com/inet-framework/inet/blob/v4.2.0/examples/wireless/nic/omnetpp.ini#L245 for an example configuration. If not, all your cars can use purely INET wireless models.