Trajectory Prediction in Campus Based on Markov Chains
In this paper, we present a model of predicting the next location of a student in campus based on Markov chains. Since the activity of a student in campus is closely related to the time at which the activity occurs, we consider the notion of time in the p
- PDF / 1,884,790 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 3 Downloads / 198 Views
Abstract. In this paper, we present a model of predicting the next location of a student in campus based on Markov chains. Since the activity of a student in campus is closely related to the time at which the activity occurs, we consider the notion of time in the prediction algorithm that we coined as Trajectory Prediction Algorithm (TPA). In order to evaluate the efficiency of our prediction model, we use our wireless data analysis system to collect real spatio-temporal trajectory data in campus for more than seven months. Experimental results show that our TPA has increased the accuracy of prediction for over 30 % than the original Markov chain. Keywords: Wireless data analysis system regularity Markov chain
Trajectory prediction
Activity
1 Introduction With the rapid development of radio access technologies, it provides much convenience for us to collect a large number of user location information. An individual carrying the mobile phone unintentionally generates many spatio-temporal trajectories that are represented by a sequence of detection device IDs with corresponding access times [1]. When the individual does not know that his movements are recorded, these trajectory information can reflect the individual’s actual activity regularity. Individual trajectory prediction plays a very important role in nowadays such as the traffic planning, urban planning and control of influenza problem. In this paper, we established a model for predicting the student’s trajectory in campus. Since the location-based service is based on the location of its user, predicting the next location of an individual can provide the recommendations of the corresponding surrounding restaurants, supermarkets or gas stations. In traffic planning, predicting activities of track traffic participants can inform possible future traffic conditions, which can further help the implementation of traffic control. In network optimization, predicting the trajectory of an individual can help the SDN controller to prepare the relevant cells before the individual arrives to guarantee seamless handover authentication, and then ensure seamless user experience during mobility [2]. On the basis of Markov process, we consider the notion of the time with student activity regularity, model for the student’s trajectories and predict the next location of the © Springer International Publishing Switzerland 2016 Y. Wang et al. (Eds.): BigCom 2016, LNCS 9784, pp. 145–154, 2016. DOI: 10.1007/978-3-319-42553-5_13
146
B. Wang et al.
student. We also calculate the accuracy and the predictability for the prediction of the next location to evaluate our forecast model. The remainder of this paper is organized as follows. First, we introduce the related work in Sect. 2. Then we describe the trajectory prediction model, the Trajectory Prediction Algorithm (TPA) for the trajectory prediction in campus and the evaluation method of TPA in Sect. 3. Afterwards, we present the experiment analysis and evaluation results with the data of campus students in Sect. 4. Finally,
Data Loading...
