Deep learning enabled dual optimization for intelligent mobility and resource handling in 6G

6G networks enable seamless connectivity with intelligent mobility and efficient resource use. However, dynamic user movement and ultra-dense base stations pose challenges in roaming prediction, mobility management, load balancing, and adaptive resource allocation. To overcome these limitations, a novel Dual Optimization driven Deep intelligent framework for mobility and resource handling in 6G networks (DOD-6G) is proposed. Trajectory extraction and Doppler-based speed estimation are used to accurately identify mobility patterns and velocity-aware user movements. The spatial–temporal features are processed using the BiLSTM technique to predict roaming probability and future user locations. Red Deer Optimization distributes roaming users to reduce base station load, while Secretary Bird Optimization enables adaptive allocation of bandwidth, power, and spectrum. The proposed approach improves the overall accuracy by 1.1%, 0.8%, and 1.8% over the existing ML, HDQL, and SpiderNet approaches.