QoS enhancement with deep learning-based interference prediction in mobile IoT

Abstract With the acceleration in mobile broadband, wireless infrastructure plays a significant role in Internet-of-Things (IoT) to ensure ubiquitous connectivity in mobile environment, making mobile IoT (mIoT) as center of attraction. Usually intelligent systems are accomplished through mIoT which demands for the increased data traffic. To meet the ever-increasing demands of mobile users, integration of small cells is a promising solution. For mIoT, small cells provide enhanced Quality-of-Service (QoS) with improved data rates. In this paper, mIoT-small cell based network in vehicular environment focusing city bus transit system is presented. However, integrating small cells in vehicles for mIoT makes resource allocation challenging because of the dynamic interference present between small cells which may impact cellular coverage and capacity negatively. This article proposes Threshold Percentage Dependent Interference Graph (TPDIG) using Deep Learning-based resource allocation algorithm for city buses mounted with moving small cells (mSCs). Long–Short Term Memory (LSTM) based neural networks are considered to predict city buses locations for interference determination between mSCs. Comparative analysis of resource allocation using TPDIG, Time Interval Dependent Interference Graph (TIDIG), and Global Positioning System Dependent Interference Graph (GPSDIG) is presented in terms of Resource Block (RB) usage and average achievable data rate of mIoT-mSC network.