Performance Evaluation of Weighted Time-Parameterized Edit?Distance based Trajectory Clustering in RFID Environment

Technologies such as Radio-Frequency Identification (RFID) are used for automatic tracking and tracing in a wide range of applications. RFIDs are accompanied by noise. The Performance of RFIDs is dependent on cleaning, filtering and reducing errors. RFID tags data capture can be enhanced using trajectories for dataset generation and applying mining, clustering and, probabilistic techniques. Trajectory clustering is using RFID data management and mining has a wide range of applications in various areas, such as traffic monitoring, video surveillance, cattle tracking and supply chain management. Trajectory clustering is the method of classifying similar trajectories according to a similarity distance. Depending on the task, given a set of trajectories, one may want to find clusters of objects that followed the same path or detect groups that moved together for given period of time. Hierarchical RFID Trajectory Clustering uses similarity measure called, Time-parameterized Edit Distance (TED) is improvised weightage for the different parameters. TED considers the same weightage for all the parameters. Our proposed weighted time parameterized similarity measure is used to find the similarity between to trajectory path by taking into consideration time dimension values in the calculation. Also, this model can deal with variants in both time and space dimensions and the clustering algorithm are much less sensitive to noise and outliers than existing methods. In this paper we are presenting performance evaluation of our proposal.