<div class="csl-bib-body">
<div class="csl-entry">Yang, J., Yang, K., Xiao, Z., Jiang, H., Xu, S., & Dustdar, S. (2023). Improving Commute Experience for Private Car Users via Blockchain-Enabled Multitask Learning. <i>IEEE Internet of Things Journal</i>, <i>10</i>(24), 21656–21669. https://doi.org/10.1109/JIOT.2023.3317639</div>
</div>
-
dc.identifier.issn
2327-4662
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/190628
-
dc.description.abstract
With deepening urbanization and Internet of Vehicles (IoV) applications, the number of private cars has been increasing in recent years. However, because the surging number of private cars is not compatible with limited road resources, private car users have had unsatisfactory commute experiences during their daily travel. In this work, we focus on improving private car users’ commute experience based on an analysis of IoV trajectory data in a privacy-preserving way. Our idea is based on the following observations: i) The commute experience of private car users is closely related to the departure time and the travel cost; ii) Most travel costs are spent on urban hot zones. Motivated by these findings, we propose a novel blockchain-enabled model named DeepICE (Deep Improving Commute Experience) to improve private car users’ commute experience by predicting when to depart and when to arrive. In this model, a blockchain with a consensus mechanism is developed to address private car user privacy concerns. In addition, we propose a multitask learning-enabled GCN (graph convolution network) method to capture the highly complex features and relations between two tasks, i.e., the departure time and travel cost, and then develop the model to predict these two tasks. The experimental results demonstrate the superior performance of our proposed model compared to existing approaches. Our model can be applied to efficiently enhance private car users’ commute experience.
en
dc.language.iso
en
-
dc.publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
-
dc.relation.ispartof
IEEE Internet of Things Journal
-
dc.subject
blockchain
en
dc.subject
commute experience
en
dc.subject
multitask learning
en
dc.subject
privacy-preserving
en
dc.subject
private car
en
dc.title
Improving Commute Experience for Private Car Users via Blockchain-Enabled Multitask Learning