While the emergence of Connected and Autonomous Vehicles (CAVs) holds great potential to improve road safety for Vulnerable Road Users (VRUs), achieving this depends on accurate location readings of both the CAVs and VRUs, particularly in scenarios where the sensors of CAVs are ineffective (i.e., sensor blind spots). Currently, the accuracy of common commercial Global Navigation Satellite System (GNSS) devices still typically ranges from meters, which proves insufficient for most vehicular safety applications. Therefore, a method to improve this accuracy is needed. In addition, GNSS accuracy varies depending on the technology used by each specific device carried by road users, making implementing GNSS accuracy enhancement at the device level less feasible. This paper proposes a centralized system leveraging Kalman Filters to enhance the accuracy of heterogeneous devices. Additionally, it presents tests on various types of Kalman Filters and compares their performance.