Virtual Reference Tag Assisted Radio Frequency Identification Localization And Tracking Using Artificial Intellect Techniques In Indoor Environment

Abstract

Radio Frequency Identification (RFID) technology is used to localize and track mobile objects using radio frequency communication in an indoor environment. Generally, the localization method is based on Received Signal Strength Indicator (RSSI) readings. However, improving the tracking accuracy and reducing tracking errors is still challenging in an RFID-based indoor environment. To overcome these problems, we proposed the VIRALTRACK (Virtual Reference Tag Localization and Tracking) model, which includes four processes: signal improvement, optimization-based virtual reference tag allocation, quantum-based localization and deep reinforcement learning-based tracking. In the first process, we proposed an Extended Gradient Filter (EGF) algorithm for removing the RSSI fluctuations to improve the efficiency of the signal. In the second process, we proposed the Emperor Penguin Colony (EPC) optimization algorithm for allocating the virtual reference tag by considering SNR, number of tags and environmental factors (temperature, humidity). Based on this information, the RFID reader allocates the virtual reference tags for each gird, increasing the tracking accuracy. In the third process, we estimate the moving target's position by performing localization using Quantum Neural Network (QNN). To choose the optimal virtual reference tag for localization, we proposed the SignRank algorithm, which reduces the errors during tracking. Finally, we proposed Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm for tracking by considering the distance, phase, orientation and precious coordinates, which effectively tracks the moving target tag, thus increasing tracking accuracy. The NS3.26 network simulator conducts the simulation, and the performance is evaluated regarding tracking accuracy, tracking error, and cumulative probability.