5G network slicing, Unmanned Aerial Vehicle (UAV) aided Multi-Access Edge Computing (MEC), and blockchain technology create new options for intelligent resource management. However, a major difficulty is efficiently allocating limited UAV resources to heterogeneous device demands while guaranteeing slice-specific Quality-of-Service (QoS) criteria. We develop a blockchain-empowered clustering framework that couples seven clustering algorithms (K-Means, DBSCAN, OPTICS, MeanShift, Hierarchical, Gaussian Mixture Model, and Divisive clustering) with an upgraded contextual bandit decision agent. The agent dynamically chooses between local execution and UAV offloading depending on slice characteristics, signal strength, latency constraints, and cluster homogeneity parameters. We use dual smart contracts (UserAllocation and UserOffload) on a permissioned Ganache (Ethereum) blockchain to ensure trust and transparency. An appropriate model of UAV placement considers the signal propagation effects and spatial distribution. The experimental results demonstrate that our clustering-aware method achieves significant improvement over non-clustered baselines by 27% in task offloading reward, 43% in regret reduction, and 20% in action accuracy, while maintaining high SLA requirements for all slice types (URLLC, eMBB, mMTC). The best overall performance is obtained with the MeanShift method. Dynamic device clustering is not only a data organization approach but also an excellent control mechanism for slice-aware resource orchestration in UAV-MEC systems. The combination of clustering intelligence, contextual bandit decisions, and blockchain-based trust gives a scalable, adaptable, and accountable solution for 5G/6G UAV-assisted edge networks.
