Cyber-physical system has been widely used in various areas as the integration of computing and physical system. As a typical application of cyber-physical system, cyber-physical surveillance system for transportation (CPSST) allows real-time video monitoring to facilitate the deploying of smart transportation systems. For video streaming in CPSST, dynamic radio spectrum management is a key technology dealing with the current situation that the spectrum resource is almost used up. In this paper, taking into account the application layer quality of service and wireless link security, a novel dynamic spectrum management scheme has been proposed to minimize the system cost of CPSST. Video distortion is considered as the application layer quality of service metric, and the system cost is defined as a combination of distortion and security cost. We use intra-refreshing rate in video coding to minimize the distortion. Furthermore, the problem is formulated as a restless bandit system, which uses current and historical information to optimize the action, with the objective of maximizing the total discounted system reward. We also describe the two spectrum management processes. Extensive simulation results are presented to demonstrate the significant performance improvement of the proposed scheme compared with the existing one that ignores video distortion and subband security optimization. Copyright © 2014 John Wiley & Sons, Ltd.