Performance Evaluation of Velocity Partitioned Moving Object Indexing in the Context of PCA and ICA

Abstract

the advent of new mobile services and access feasibility, moving object indexing has grabbed a great number of new research techniques and challenges. But the data exploitation in this area is based on the straitened or symmetric data. The real world data is multi-dimensional and the objects like GPS enabled devices, flights and vehicles on road networks would be skewed in nature. Velocity of an object is of utmost importance as these objects are dynamic. Velocity Partitioning (VP) is a method to cater the need and to support the real world multi-dimensional data. This technique not only improves the query processing but also indexes the moving objects efficiently. In this paper, Velocity Partitioning technique is adopted and Independent Component Analysis (ICA) technique is used instead of Principle Component Analysis to find velocity axis. In the experimental section, different real-time and synthetic datasets are considered; Indexed the moving objects in both VP with PCA and ICA cases. Indexing structures like R*-tree and B + -trees are used for making the analysis. The performance of state-of-the-art indexing structures like B x and TPR* trees can be improved if VP is applied afore.