Sensing data gathering and dissemination is one of the most challenging tasks in smart city construction, and vehicles moving around a city have been widely considered as a good candidate to deliver data efficiently and economically. Hence, this paper proposes a deep recurrent neural network-based algorithm to predict vehicle mobility and facilitate vehicle-based sensing data delivery. Extensive evaluations have been conducted by using a large-scale taxi mobility dataset that is obtained from a smart city testbed deployed in Tokyo, Japan. The results have validated that, compared with the most state-of-art algorithms, our proposal can improve the F1-Score of vehicle mobility prediction by a range of 18.3% ~24.6%.
Applying Deep Recurrent Neural Network to Predict Vehicle Mobility
2018-12-01
510823 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Multivariate Time Series Analysis Using Recurrent Neural Network to Predict Bike-Sharing Demand
Springer Verlag | 2020
|