Authors: J. Vinoli, B.Sri Chithra Devi, A.Ambika, L.P.Sujitha
Abstract: The advent of the Internet of Things (IoT) technology has led to many transformative changes in retail inventory management. This has been achieved through the resolution of longstanding problems, including stockouts, overstocking and inefficiency in operations. This paper presents a robust IoT-based framework for smart inventory management which includes the use of Radio-Frequency Identification (RFID), Bluetooth Low Energy (BLE) beacons, weight sensors and machine learning algorithms for predictive analytics in conjunction with edge computing. The proposed approach provides real-time visibility on stock, automatic alert notifications of stock replenishments and demand forecasting with 99.35% accuracy in identifying products and response time of 8.66 seconds for notification delivery. Comparative study findings reveal that IoT-based systems provide inventory accuracy of more than 95% as opposed to 70-75% in case of manual and barcode-based systems. The proposed framework minimizes stockout rate by 23.5% and forecast error by 18.2% using multi-agent deep reinforcement learning technology.
DOI: http://doi.org/10.5281/zenodo.21676460
