A Capacity Analysis for the Transmission of Event and Cooperative Awareness Messages in LTE Networks

This paper presents the results of a performance study on cooperative vehicle services in Long Term Evolution (LTE) networks. By means of radio network simulations, the system capacity and its limiting factors are analyzed for the message types defined by ETSI ITS. For the Cooperative Awareness Message (CAM), the system is limited by the downlink data channel capacity. For the standard scenarios with a 10 Hz sending rate, even a distribution to vehicles in the immediate vicinity can fully load an LTE network. This effect is a consequence of the CAM traffic pattern, where each vehicle continuously sends data which is distributed to all neighboring vehicles. For the simulated LTE unicast connections from the network to the vehicles, this means that the downlink traffic increases with the number of neighbors and is therefore a multiple of the uplink traffic. Although the supported number of vehicles per cell can be increased by strict filtering or lowering of the sending rate, with the exception of special use cases this type of message application results in an extremely high traffic load and seems unsuitable for the transmission over cellular networks as investigated in this paper. However, similar issues with high traffic load also seem to exist in ad-hoc networks and relevant standardization bodies such as ETSI have started to investigate congestion control mechanisms, so that this assessment may need to be reviewed in the future. Event-related messages such as the Decentralized Environmental Notification Message (DENM) generally have a significantly higher information value than CAMs because they are only sent when a concrete condition has been observed. This makes it easier to justify using network capacity to transmit these messages. For DENMs, the number of senders is significantly lower, and the network behavior is stable for typical scenarios. Therefore, the transmission of DENMs over LTE is judged feasible for the relevant use cases. Also, in contrast to the CAM traffic pattern, congestion is a temporary effect with a limited lifetime such that the network load decreases again when the event ends.