I. Introduction
The emergence of the Internet of Vehicles (IoV) has enabled ordinary vehicles and roadside units units (RSUs) to have computing and communication capabilities; at the same time, intelligent transportation systems (ITSs) have become possible. In the increasingly intelligent IoV, how to ensure the stable operation of various software to maintain traffic safety and network stability is very important [1]. In [1],the concept of the Cognitive Internet of Vehicles (CIoV) was proposed to help solve the problems of cellular networks, and vehicle self-organizing networks, which cannot effectively guarantee appropriate costs and stable connectivity. Recently, CIoV has attracted the attention of scholars in many fields, such as autonomous driving and ITS [2]–[4]. However, with the increase in the services deployed in the IoV, the maintenance and update of these software itself become a very important issues [5]. To enable CIoV to have the ability to perceive and evolve while having cognitive capabilities, we studied software escalation prediction in CIoV.