Non-Intrusive Cell Load Estimation in Commercial LTE Networks for Media Selection in Hybrid V2X Systems
With the advent of intelligent transportation systems and their associated more stringent requirements in terms of latency and packet delivery ratio, existing radio access technologies have exhibited a limited performance in specific scenarios that makes their sole operation in isolation unreliable for safety-related applications. In this context, hybrid V2X systems represent the state-of-the-art approach to address such challenge. Under this paradigm, the efficient operation of the system relies on a media selection algorithm, which uses different network parameters for predicting the technology that is foreseen to perform the best in a given scenario. While the network load is here regarded to affect the latency of the communication, no quantitative study has been found in the literature which supports this claim. Furthermore, the inaccessibility to this metric for commercially available off-the-shelf user equipments prevents a straightforward analysis of such parameter and its effects on C-V2X communications. In order to fill that gap, a framework has been designed and implemented over the course of this thesis that non-intrusively gathers information from live LTE networks when traversing them using an ITS-enabled vehicle. Such information has been then post-processed and used to perform a mapping of the deployed frequency bands in both rural and urban environments, as well as for evaluating principally the relationship between the latency experienced in on-road LTE communications with the instantaneous uplink load estimated for the serving eNodeBs traversed in an urban environment, without any assistance of the network. Thus, the overall uplink load has been estimated and in-depth analysis on the load per eNodeB has been carried out; during the tests, however, the uplink load has been estimated to be below 30% for the scenario under analysis and no straightforward correlation with the latency experienced when transmitting messages with different sizes has been identified, which suggests that further measurements need to be taken. On the other side, it was found that the radio resource allocation scheme used for uplink grant assignments by the eNodeBs under analysis is exploitable for reducing the cell load estimation time, which will support the migration of the set-up towards a real-time realization. The developed system therefore constitutes the groundwork for later analysis that will allow a more straightforward characterization of this relationship in different scenarios, which will ultimately serve to improve the media selection algorithm in hybrid V2X networks.
Aachen, TH, Master Thesis, 2018