Abstract Handover (HO) triggering in fifth-generation (5G) ultra-dense networks (UDNs) is critical for ensuring seamless connectivity as mobile user equipment (UE) transitions between adjacent small cells or base stations. Due to the dense deployment of low-powered cells and high user mobility, UDNs experience frequent handovers, which can lead to service degradation and reduced quality of service (QoS). Efficient and adaptive handover mechanisms are therefore essential for sustaining network performance. This study proposes a hybrid intelligent framework for adaptive handover triggering in hyper-dense 5G scenarios. The framework integrates a modified pelican optimization (MPO) algorithm for efficient user clustering, a hybrid quantum–classical recurrent neural network (QCRNN) for dynamic handover decision-making, and chaos gorilla troops optimization (CGTO) for predictive mobility pattern analysis using historical user data. The QCRNN leverages design constraints including transmission delay, signal-to-interference-plus-noise ratio (SINR), received signal strength (RSS), user motion potential (UMP), and current load conditions (CLC) to determine optimal handover initiation without predefined thresholds, enabling adaptive and context-aware decisions. Simulation results demonstrate that the proposed MPO–QCRNN–CGTO approach significantly outperforms existing methods (RSRP, fuzzy, AHP–TOPSIS–Q, FMCSS, SC-Q). The framework reduces the average number of handovers (NOH) by up to 96%, decreases the probability of ping-pong handovers (PPHO) by up to 87%, and lowers handover failure rates by up to 78%. Furthermore, it improves throughput by up to 287% and reduces network latency by up to 65% across varying user densities and simulation times. These improvements highlight the framework’s ability to minimize unnecessary handovers, prevent service interruptions, and maintain high network efficiency, confirming the effectiveness of the integrated MPO–QCRNN–CGTO approach in enhancing mobility robustness and QoS in 5G UDNs.
Rajesh et al. (Fri,) studied this question.