Optimizing IoT Service Matching Using Simulated Annealing Enhanced K-means Clustering

With the wide application of IoT technology, traditional IoT service models face challenges in terms of service matching efficiency and computational burden. Although the existing K-means clustering algorithm is widely used, it is sensitive to the initial centre of mass selection and is prone to fall into local optimal solutions, which affects the accuracy of the clustering results and the service matching efficiency. For this reason, the study proposes a K-means clustering method combined with a simulated annealing algorithm. By simulating the physical annealing process, the local optimal problem is effectively avoided and the global optimization ability of clustering is improved. The experimental results show that the Silhouette score of SA-K-means clustering is 0.82, and the Davis Boulding index is only 0.41. Under high concurrency, the proposed algorithm achieves a data reception accuracy of 97.8%, a response time of 1276.18ms, and a service success rate and reliability of 95.12% and 96.75%. In addition, the median inverse generative distance of SA-K-means in different complexity scenarios is 0.0061 and 0.0065, which is closest to the Pareto optimal frontier. This study provides a new theoretical approach for IoT service matching, enriches the research content, and provides a reference for other fields that require efficient clustering processing.