Modeling and Prediction of Vehicle Routes Based on Hidden Markov Model

Autor(es)

Akabane, Ademar T. ; Pazzi, Richard W. ; Madeira, Edmundo R. M. ; Villas, Leandro A. ; IEEE

• Tipo de documento: Artigo Científico
• Fonte: 2017 IEEE 86TH VEHICULAR TECHNOLOGY CONFERENCE (VTC-FALL); v. N/A, p. 5-pg., 2017-01-01.

Resumo

Understanding traffic conditions, in an urban environment, by means monitoring or/and predicting is not an easy task. In Intelligent Transportation Systems, the reliable vehicle route prediction has meaningful application value. Vehicle route prediction can increase the variety of VANETs applications such as predicting traffic situation ahead, optimal route recommendation, driver assistant, and automatic vehicle behaviors. Due to the challenge imposed by the vehicle route prediction and its wide application, researchers in both industry and academia have focused their efforts on this area. In order to explore this area, this paper describes an approach to predict the vehicle's future path in a realistic urban scenario. For that, it employs a hidden Markov model along with the outcome of the Viterbi algorithm to make a probabilistic prediction. The parameters modeling is estimated based on information extracted from the travel route dataset and computed in an offline manner. In the online phase the routes prediction is carried out. Our approach has high accuracy rate according to the numerical simulations results. Moreover, we believe that the VANETs applications previously mentioned can take advantage of our approach. (AU)

• Processo FAPESP: 16/24454-9 - Agregação de Dados em VANETs
• Beneficiário: Ademar Takeo Akabane
• Modalidade de apoio: Bolsas no Exterior - Estágio de Pesquisa - Doutorado
• Processo FAPESP: 15/25588-6 - Gerenciamento distribuído de informação em redes sociais veiculares
• Beneficiário: Ademar Takeo Akabane
• Modalidade de apoio: Bolsas no Brasil - Doutorado