Using Spatial Trajectory Prediction to Locate Mobile Objects

Inertia has a moving object follow a path or trajectory that resists any change in its motion. Human travel patterns normally have the similar inertia feature. For example, the vehicles on a highway usually stay on the highway or people tend to walk towards a popular destination such as a mall or a park. This research tries to find the anticipated locations/paths of moving objects by using spatial trajectory prediction. One example of this research is as follows. Assume you have several friends who are supplying their mobile locations to you constantly. Now, you want to reach any one of them, but the problem is their locations are dynamic instead of static. Using a method of spatial trajectory prediction, we may predict our friends' forthcoming locations and find the ones who are close to us in the next moment.

Global positioning system (GPS), handheld/mobile/smartphone computing, location-based services (LBS), human behavior recognition, social networks, spatial trajectories

The Proposed Method I
This research tries to locate mobile objects by using spatial trajectory prediction. The proposed method includes the following five steps:
Location data collection
Mobile users' locations (including longitudes, latitudes, and times) need to be sent to the system from time to time. In order to do this, the mobile users need to install the app in their devices.

Location data preparation
Location data supplied by the GPS (Global Positioning System) is usually not stable and may contain noises. Before it is used by the system, the location data must be processed.

Location data storage and indexing
The amount of location data collected could be huge because many locations of numerous objects need to be saved. Additionally, the vast amount of location data makes the indexing and searching more difficult. Innovative methods have to be used to store, index, and search location data.

Spatial trajectory mining and prediction
Route projection is used to predict the future trajectory. However, this approach suffers from a problem of unrealistic prediction. It is because the prediction is based on route projection and the projection may not have a road. I haven't found the algorithms yet, but it should not be difficult if the following steps are taken. First, list all kinds of trajectory prediction. For example,
    --->--> prediction: --->-->-->
I can list at least 10 cases. Study these cases and come up with an algorithm.

Using spatial trajectory prediction to locate moving objects
Once the predicted locations are found, finding the nearest objects becomes a trivial task. Euclidean metric can be used to find the distance.

The Proposed Method II
Collect and save all spatial trajectories. Convert the trajectories into a finite automaton. When a trajectory prediction is needed, use the finite automaton to find it. This method will guarantee the prediction includes roads. The problem of this approach is it is not very innovative.

It is believed that the number of smartphones sold will surpass the number of plain mobile phones sold in the near future. Compared to plain mobile phones, smartphones are able to perform many more advanced functions such as mobile Web browsing, mobile office, and mobile gaming. One of the mobile applications, location-based services (LBS), has attracted great attention recently. A location-based service is a service based on the geographical position of a mobile handheld device. This research proposes location-based research, which uses spatial trajectory prediction to locate mobile objects, whose locations are constantly changed. Preliminary experiment results show the proposed methods are effective and easy-to-use.

  1. Steiniger, S., Neun, M., & Edwardes, A. (2006). Foundations of Location-Based Services. Retrieved May 13, 2009, from
  2. Zheng, Y. & Zhou, X. (2011). Computing with Spatial Trajectories, Springer.