Data Classification and Transportation in Rail Networks

Publication Type:
Thesis
Issue Date:
2020
Full metadata record
IoT is a revolutionary technology in the digital world, with a diverse range of services being created and deployed. One of the major challenges involved in efficiently implementing IoT is the management and transportation of large volumes of data that this solution generates. Modern approaches for IoT completely rely on cellular networks. As the demand for such networks is massively growing, in this thesis, we explore other communication methods as alternatives for management and delivery of IoT data in rail networks. Particularly, the focus will be on developing strategies that utilize existing trains and the rail network as a mode of data transportation. Furthermore, the thesis will combine physical delivery of IoT data by trains to strategic collection points in rail networks with cellular infrastructure to minimize costs and increase communication scalability and efficiency. Therefore, in this thesis, we introduce a new framework into future data-driven rail networks. For this purpose, we propose an edge processing unit that includes two main parts. The first part is a data classification model that classifies IoT data into maintenance-critical data (MCD) and maintenance-non-critical data (MnCD). The second part is a data transmission unit that based on the class of data, employs appropriate communication methods to transmit data to strategic collection points. The MCD is immediately forwarded through real-time communication methods such as cellular networks. However, for the transmission of MnCD, we propose three travel pattern methods including train-to-station (T2S), train-to-train (T2T) and train-to-wayside (T2W) communications that employ trains as data carriers. We validate the classification model and all the transmission methods through extensive experiments. The simulation results show the effectiveness of our models as follows. The data classification model was validated under different operating conditions with over 98% accuracy. For the T2S model, we showed that over 5 GB data can be offloaded through T2S communications. Additionally, our proposed mobility model for T2T communications was tested with real GPS data and showed over 98% accuracy. Furthermore, for the T2W communications, we showed that the proposed AP placement approach could improve the efficiency of data offloading up to 165%. Finally, we proved that we can offload over 250 Gigabits through T2W communications over WiFi networks.
Please use this identifier to cite or link to this item: