Resource allocation in wireless networks is an important factor as it defines e the utilization of spectrum usage, resource distribution, and quality of services. The evolution of mobile communication brings additional challenges in allocating the resources due to high user mobility, heterogenous traffic demands, and dynamic topologies. Conventional techniques lag in performance due to their static optimization procedures and limited spatial-temporal awareness. To overcome this, a Spatio-Attentive Graph Mixture Network (SAGMNet) is proposed in this research work for enhanced resource management. The proposed model incorporates graph-based learning with a multi-modal attention mechanism for feature processing and scheduling decisions. The experimental analysis of the proposed model utilizes benchmark vehicular wireless scheduling dataset and evaluates the model's performance with different metrics like spectrum utilization, throughput, and latency. The proposed model exhibits superior performance in terms of 93.6% spectrum utilization efficiency, 29.1 Mbps average throughput, 0.087 interference index, 3.26 Mbps/Watt energy efficiency, 0.961 scheduling fairness, 5.9ms allocation latency, 0.928 mobility robustness score, and 3.2 ms inference time, which is better than conventional DNN, GCN, LSTM, ST-GCN, and Transformer-GAT models.

Context-Aware Scheduling Spatio-Temporal Graph Learning Adaptive Resource Management Intelligent Wireless Networks Dynamic Topology Optimization Mobility-Robust Allocation