I have a large trajectory dataset generated by cars' GPS. Since I will do some data mining work on them, I need an efficient way to store and index them. I read some papers about Trajectory Indexing (Computing with Spatial Trajectories, by Yu Zheng and Xiaofeng Zhou, Microsoft research Asia), 3D R-tree, TB-tree(Trajectory-Bundle tree), SETI(Scalable and Efficient Trajectory Index) were proposed to solve this problem. But it is quite difficult to implement them. Is there any implementation on this problem?
I am not quite familiar with POSTGIS. It has R-tree, but trajectories have a "time" dimension. Could I expand postGis to index trajectories? How?
Truely grateful for all anyone's help!

1,2000 taxis' trajectories for one month. About 50GB. latitude,longtitude,time. I need to search for trajectories of regions during anytime, and so on.

  • Welcome to gis.SE. Can you add some more specifics, such as "how large", whether you have 2D+time, 3D+time or something else you need to search on, and what sort of data mining you need to do? You can just click edit below your question to add this information. – BradHards Dec 28 '13 at 9:39
  • This is a really good question, I have already read the same papers and techniques and I have the same question. But here is stackexchange. Lets close questions because aren't fit with the standards of the site. – Telperinquar Nov 5 '18 at 14:18