Skip to Main Content
Radio Frequency Identification (RFID) can be used in various ways for the optimization of supply chain management processes. However, there are technological constraints that delay a reliable and productive use of the technology. One of these constraints is the problem of false-positive RFID tag reads i.e., tags that have been read unintentionally by an RFID reader. We propose a machine learning based approach that makes use of the low-level reader data collected when reading tags to detect such false-positives. We evaluate our approach by verifying it with data collected in a productive RFID enabled distribution center, where it is necessary to distinguish between pallets that are loaded onto trucks and pallets that are in range of the reader by accident only. Furthermore, we identify several attributes which are expected to reveal characteristics within the low-level reader data that is typical to such false-positive reads.