Download PDFOpen PDF in browser

Detecting Transportation Modes with low-power-consumption sensors using Recurrent Neural Network

EasyChair Preprint no. 492

8 pagesDate: September 8, 2018

Abstract

With the quick development of mobile Internet and the popularity of smartphones, smartphone-based transportation mode detection has become a hot topic, which is able to provide effective data support for urban planning and traffic management. Though the popular GPS based transportation mode detection method has achieved reasonable accuracy, this method consumes large power, thus limiting it to be used in smartphones. Here, we propose a novel transportation mode detection algorithm using the recurrent neural network. In order to identify transportation modes with low power consumption, this algorithm only uses four low-power-consumption sensors (namely accelerator, gyroscope, magnetometer, and barometer) which are embedded in the commodity smartphones. Furthermore, we exploited the good representative ability of Long Short-Term Memory (LSTM) and applied it to recognizing the transportation modes to achieve higher accuracy. To filter noises, a preprocessing is applied. After calculating features, we adopt the LSTM learning algorithm to train a model of transportation mode recognition and employ this model to predict transportation modes. Extensive experimental results indicate that our proposed approach outperforms the compared state-of-the-art transportation recognition methods with 96.9% accuracy to detect four transportation modes, namely buses, cars, subways, and trains.

Keyphrases: deep learning, Recurrent Neural Network, transportation mode detection

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@Booklet{EasyChair:492,
  author = {Hao Wang and Haiyong Luo and Fang Zhao and Yanjun Qin and Zhongliang Zhao and Yiqu Chen},
  title = {Detecting Transportation Modes with low-power-consumption sensors using Recurrent Neural Network},
  howpublished = {EasyChair Preprint no. 492},
  doi = {10.29007/hwnb},
  year = {EasyChair, 2018}}
Download PDFOpen PDF in browser