Implementation and experimental evaluation of a Local Dynamic Map for Road Data Aggregation with the support of vehicular micro-clouds

The concept of autonomous driving in today worlds’ transportation systems is of high importance. A lot of studies are investigating the challenges and opportunities related to the transportation policies that will arise as a result of emerging Autonomous Vehicle (AV) technologies. In addition to Intelligent Transport Systems (ITS) applications that rely only on the sensor information of an ego vehicle, recent studies are also examining cooperative ITS paradigms on which the sensor information is shared through by the use of the vehicle-to-everything (V2X) communications. This information could be exchanged in the network based on standardized ITS messages, defined by the European Telecommunications Standards Institute (ETSI), more specifically Cooperative Awareness Message (CAM), Decentralized Environmental Notification Message (DENM) and Collective Perception Message (CPM). One of the key facility elements in the co-operative ITS, is the Local Dynamic Map (LDM) that maintains the information about all the objects influencing or being part of the traffic, enabling ITS applications to retrieve it on demand. All the V2X-enabled vehicles have an LDM, which is a database filled with information from V2X messages such as CPMs and CAMs coming from nearby vehicles and the on-board sensors of the ego vehicle. The information found in CAMs are about the vehicles that are sending the message and the CPMs, instead, contain information about the vehicle’s detected objects, found inside its LDM. For a group of neighboring vehicles, because of the repetition of the gathered context information, there is the possibility to create a cluster, often referred to as vehicular micro-clouds, which enables the possibility to collectively process and aggregate the data. Platooning is one of the most popular V2X-enabled applications suitable for this architecture to be deployed in real life. Platoons are a cluster of vehicles that manage to travel closely behind each other, in order to save fuel, reduce traffic and increase road safety. However, locally matching and aggregating the data coming from other vehicles periodically is computationally expensive. To overcome this issue the solution on which this thesis puts is basis, is the Platoon Local Dynamic Map (P-LDM), which is a single aggregated dataset of information for the whole platoon. In this service, each platoon member will share its LDM information with the platoon leader, which in return will assign a subset of the platoon database to the member to aggregate. This solution not only aims to solve the sensing limitation of the platoon members, hence improving the safety of the platooning maneuvers, but also decreases the computational load on a single member. Finally, the core of this work has been to evaluate the collective perception enhancement and the distribution of the computational load among the platoon members, to this end, we make use of MS-VAN3T, an open-source vehicular network simulation framework able to integrate multiple communication stacks such as IEEE 802.11p or LTE-V2X, built on the ns-3 simulator and the Simulation of the Urban Mobility (SUMO) to manage the mobility. In MS-VAN3T the message exchange is based on the standardized ITS messages defined by ETSI and It gives us the opportunity to model all aspects of communication among the various entities in a created vehicular scenario.