Exploring Trajectory Prediction Through Machine Learning Methods

Rongpeng Li,Honggang Zhang,Tariq S. Durrani,Lin Ma,Chujie Wang

doi:10.1109/access.2019.2929430

Rongpeng Li, Honggang Zhang + Show 3 more

Open Access

https://doi.org/10.1109/access.2019.2929430

Copy DOI

Journal: IEEE Access	Publication Date: Jan 1, 2019
Citations: 81	License type: CC BY 4.0

Affiliation: Zhejiang University

Abstract

Human mobility prediction is of great importance in a wide range of modern applications in different fields such as personalized recommendation systems, the fifth-generation (5G) mobile communication systems, and so on. Generally, the prediction goal varies from different application scenarios. For the applications of 5G network including resource allocation and mobility management, it is essential to predict the positions of mobile users in the near future from dozens of seconds to a few minutes so as to make preparation in advance, which is actually a trajectory prediction problem. In this paper, with the particular focus on multi-user multi-step trajectory prediction, we first design a basic deep learning-based prediction framework, where the long short-term memory (LSTM) network is directly applied as the most critical component to learn user-specific mobility pattern from the user's historical trajectories and predict his/her movement trends in the future. Motivated by the related findings after testifying and analyzing this basic framework on a model-based dataset, we extend it to a region-oriented prediction scheme and propose a multi-user multi-step trajectory prediction framework by further incorporating the sequence-to-sequence (Seq2Seq) learning. The experimental results on a realistic dataset demonstrate that the proposed framework has significant improvements on generalization ability and reduces error-accumulation effect for multi-step prediction.

Highlights

Increasing pervasive usage of smart-phones and locationbased services around the world has contributed to vast and rapid growth in mobility data
Human mobility prediction is of great importance in a wide range of modern applications, ranging from personalized recommendation systems to intelligent transportation, urban planning, and mobility management in the fifth-generation (5G) mobile communication system [1], [2]
MOBILITY PREDICTION RESULTS In this part, we evaluate the performance of the proposed multi-user multi-step prediction framework on a realistic dataset

Summary

INTRODUCTION

Increasing pervasive usage of smart-phones and locationbased services around the world has contributed to vast and rapid growth in mobility data. Researchers have proposed many mobility prediction methods, such as frequent patterns mining [3], [4], Markov-based models [5], [6] and other machine learning methods [7], most of these methods are dedicated to discrete location prediction, which is a multi-classification problem, and not suitable for predicting trajectories with fixed sampling time intervals. Wang et al.: Exploring Trajectory Prediction Through Machine Learning Methods locations may have a mutation between two adjacent timesteps when the sampling time interval is large They can hardly reflect user movement trends effectively. In order to avoid the above problems, this paper takes comprehensive investigation for the approaches to predict trajectories composed of continuous coordinates Since it is a time series regression prediction problem, conventional regression algorithms such as linear regression [8] and support vector regression (SVR) [9] are candidate solutions.

RELATED WORK

Findings

CONCLUSION AND FUTURE WORKS