Abstract
One of the objectives of Intelligent Transportation Systems (ITS) is providing safety to its users. This can be accomplished by the exchange of information among vehicles, forming a network called Vehicular Ad Hoc Network (VANET). However, such information exchange is a challenge in itself. Safety messages must be timely disseminated throughout vehicles in the vicinity of a detected hazard. The easiest way to do so is by broadcasting and re-broadcasting a safety message until it reaches the whole network. Nonetheless, two problems arise from this approach: (i) the flood of transmitted messages causes massive packet collisions, incurring delays instead of quickly reaching all network nodes — an effect known as broadcast storm — and (ii) communication and processing resources are wasted because a hazard event is not necessarily required to be informed immediately to far away vehicles — they would discard the message because it does not have value for taking an action yet.
Several proposals appeared in the literature to tackle the aforementioned problems. Some techniques are based on limiting the number of re-broadcasts of a message by using counters, or by allowing retransmissions only by the vehicles farthest from the original transmitter, or even by making all vehicles share their location in order for the best forwarders to be identified. Even though these methods have performed well on highways, their use did not yield good results on urban scenarios. The higher presence of obstacles blocks the highly directive signals used for inter-vehicle communication. Therefore, another class of proposals emerged, making use of forwarder vehicles to avoid such obstacles. Still, some proposals tried to take the opportunity presented by the wide coverage provided by cellular networks, however up to LTE, communication delays were higher than the limit allowed by safety messages dissemination. Nonetheless, the development of 5G shows evidence that the technology will be able to comply with such tight delay requirements. Thus, the use of both forwarders and 5G technology for the dissemination of safety warning messages seems promising.
In this work, clustering is considered as a means to keep broadcast storms under control by limiting communication from cluster members only to the cluster master. However, creating and maintaining clusters might become a challenge because differences in the vehicles speeds and routes constantly change clusters topologies. Therefore, a mechanism must be in place to not only form clusters that will last longer, but also to merge small clusters together, since having many clusters approaches the network efficiency to the scenario without clusters. Hence, a cluster merging algorithm is proposed and, then, this work proceeds to analyze the use of different combinations of forwarders and the 5G infrastructure to achieve the best warning message dissemination performance in terms of delay and percentage of informed vehicles. In that regard, this work will follow a simulation approach to assess the mentioned performance metrics of the original proposals and of other proposals selected from the literature as benchmarks.