Massive Machine Type Communications (mMTC) is a key scenario in 5G networks to support one million terminals per square kilometer within a base station’ s coverage. To enable this scenario, a clustering-based approach that partitions the terminals within the base station’s coverage area into multiple clusters has been proposed, wherein each cluster is assigned a gateway device to relay communications between terminals and the base station. Among the various clustering techniques, K-means clustering is widely adopted due to its computational efficiency and straightforward implementation. However, it suffers from an imbalance in the number of IoT devices assigned to each cluster, leading to unequal load distribution across gateways. To address this issue, this paper presents an improvement to the k-means method, introducing the SameSizeKMeans (S-Kmeans) algorithm, which equalizes the number of IoT devices per cluster while optimizing energy-efficient transmissions to the cluster center. Furthermore, Genetic Algorithms (GA) are leveraged to optimize cluster assignments and improve fairness by incorporating evolutionary strategies that adjust device distribution across clusters dynamically.