Résumé : Connected and Automated Vehicles (CAVs) rely on the correctness of position and other vehicle kinematics information to fulfill various driving tasks such as vehicle following, lane change, and collision avoidance. However, a malicious vehicle may send false sensor information to the other vehicles intentionally or unintentionally, which may cause traffic inconvenience or loss of human lives. Here, we take the advantage of cloud-computing and increase the resilience of CAVs to malicious vehicles by assuming each vehicle shares its local sensor information with other vehicles to create information redundancy on the cloud side. We exploit this redundancy and propose a sensor fusion algorithm for the cloud, capable of providing a robust state estimation of all vehicles in the cloud under the condition that the number of malicious information is sufficiently small. Using the proposed estimator, we provide an algorithm for isolating malicious vehicles. We use numerical examples to illustrate the effectiveness of our methods.