Document Type : Research

Authors

• Hamed Alizadeh ¹
• Mohammad Mofarreh ²

¹ Electrical Engineering Department, Faculty of Engineering, Azarbaijan Shahid Madani University

² University of Bonab

Abstract

In heterogeneous networks integrating Radio Frequency (RF) and Visible Light Communication (VLC) modalities, the handover process is critical for maintaining connectivity and quality-of-service (QoS) in dynamic indoor environments. This paper proposes a novel deep learning assisted heuristic algorithm (DLHA) that predicts, optimizes, and executes handover decisions in an RF-VLC integrated network. By leveraging a gated recurrent unit (GRU)-based deep neural network (DNN) to forecast channel conditions and user mobility, and coupling these predictions with a heuristic decision framework, the algorithm minimizes handover latency, reduces packet loss, and balances network load. The problem is formulated as a multi-objective optimization problem with constraints on delay, energy consumption, and interference, and is further refined using principles from optimization and Markov decision processes (MDPs). Simulation results, validated on realistic indoor channel models and mobility scenarios, demonstrate that the proposed DLHA significantly outperforms conventional threshold-based methods.

Keywords

• Handover
• RF-VLC
• deep learning
• optimization
• heterogeneous networks