PRTS: A Passive RFID Real-Time Tracking System Under the Conditions of Sparse Measurements

In many logistics and manufacturing applications, tracking of moving objects with radio frequency identification (RFID) tags on a conveyor belt is a premise for many other processes, e.g., sorting procedures, stamping IDs in a surveillance video. However, in complex industrial environments, a sharp decline in Tag Read Record (TRR) often results in severe spatial ambiguity. In this type of scenarios, existing systems cannot work effectively owing to the prevalent existence of noise. This paper proposes a Passive RFID Real-time Tracking System (PRTS) with tolerance of a small TRR, which is designed for tracking RFID-tagged mobile objects. We use detailed deduction to convert the tracking problem into a sparse signal reconstruction one. To solve this problem, we devise a novel normal sparse signal reconstruction method based on greedy pursuit by making the best of the available prior knowledge and further ameliorate it via calibration of the phase deviation from frequency and angle-of-arrival responses. Furthermore, we leverage the simplified particle filters to facilitate the real-time tracking of mobile objects on conveyor belts. We implement the prototype PRTS with commercial-off-the-shelf RFID devices and evaluate it in various scenarios. Experimental results demonstrate that PRTS achieves a mean relative error of 7 cm under the conditions of extremely sparse measurements.