Outlier Detection in RFID Supply Chain for Path Deviation Using TRAJODBSCAN

Supply chain process is one of the area of research especially when the whole process is automated through radio frequency identification (RFID) method with readers to read the tags on objects automatically and thus generating too much of raw data in the form of tag_id, location id (latitude, longitude) and timestamp of read. Mining such data is a challenging problem. For each product or group of products, the schedule is planned by the stakeholders that include suppliers, manufacturers, distributors and retailers. From suppliers to retailers many routes in the supply chain can be generated. These routes are known as trajectories due to element of time and location read by readers. The main aim of this paper is to find outlier trajectories by clustering the scheduled or planned trajectories of the objects with the current trajectory. Any deviation or abnormal path/trajectory will be the outlier one. A modified DBSCAN-based approach is used by the authors with accuracy upto 91.2% to find out the outlier trajectory as compared with other clustering-based outlier detection algorithms and thus helping in alarming the stakeholders to take further action in the supply chain process. The time taken is relatively more as compared with other algorithms but this can be easily managed by parallel and distributed systems with advance configuration.