Implementation and evaluation of a centralized Local Dynamic Map for cooperative and connected vehicles

Connected and automated vehicles represent one of the main technological advances for the automotive field, with the aim of improving safety and human daily services. In this work the current state of the art about the related most used technologies is analyzed, giving informations about the hardware architecture and the communication modes too (V2V/V2I/V2N). The object of the thesis is to explain the Server Local Dynamic Map (S-LDM) development as part of the 5G-CARMEN project with a performance evaluation related to the main modules composing this novel automotive services. The S-LDM is a 5G-enabled Mobile Edge Computing (MEC) service storing a centralized local dynamic map of the road, to which the vehicles continuously uploads data using ETSI ITS-G5 standardized messages which are delivered thanks to the AMQP messaging protocol. The S-LDM is also able to detect when a specific condition or request occurs. In this case a general context around a vehicle of interest is created and then made available to other MEC services which can control automated maneuvers in a centralized way. This thesis focused in particular on the development of a geographical filter to retrieve the messages coming from vehicles travelling on specific stretches of the road and on the performance evaluation of the S-LDM thanks to an open source vehicle simulation and emulation framework, called ms-van3t. Then each S-LDM submodule behaviour is analyzed as a function of the number of vehicles. At the end the tests results with emulated vehicles have been matched with a road test using two cars with C-V2X technology abroad. Results indicate the goodness of the S-LDM in managing several messages, the ability of the developed geographical filter to select the proper vehicular messages, and the capability of the S-LDM of maintaining a constant performance up to 25 vehicles.