Edge computing-based Internet of Vehicles (EC-IoV) has emerged when computing environment is extended to vehicular network edge for supporting compute-intensive applications. To implement EC-IoV, roadside units (RSUs) are enhanced and upgraded toward edge computing infrastructures with hardware improvements. RSU is a crucial component to communicate with and serve vehicles in EC-IoV. However, RSUs easily suffer from potential network attacks. The attacks lower the quality of user experience, disrupt RSU workload scheduling and even result in user information leakage. Due to the open deployment environment, network traffic among all RSUs may be eavesdropped by a global adversary. When a hotspot phenomenon causes heavy traffic to an RSU, the adversary could take advantage of traffic statistics to locate the RSU. After that, the adversary may concentrate to attack the RSU and try to interrupt on-going services. As a consequence, RSU hotspot attack with traffic eavesdropping is launched. In this paper, we focus on such a new adversary model in EC-IoV and discuss the attack and its defence. To prevent from the attack, we propose a proactive scheme by generating dummy traffic delivery. The local vehicles are collaborative and encouraged to send dummy packets to specified RSUs. This dummy traffic will mislead the traffic statistics and protect hospot RSUs. Stackelberg game approach is used to build an incentive mechanism in the scheme. Numerical results demonstrate that our scheme is effective and efficient to secure EC-IoV against RSU hotspot attack.