Weighted-based optimized cluster head selection and hybrid adaptive clustering approach in mobile ad-hoc networks

Mobile ad-hoc networks (MANETs) are dynamic and self-organizing wireless networks in which nodes move randomly without relying on centralized routing infrastructure. One of the major challenges in such networks is the increased energy consumption of participating nodes, which shortens the overall network lifetime and leads to instability in packet delivery. Therefore, an effective clustering strategy in MANETs is critically important for sustaining network performance. In this thesis, a Hybrid Adaptive Clustering Algorithm for Dynamic MANETs (DMHAKA) is proposed to enhance performance in dynamic MANET environments. The algorithm is designed in two stages. In the first stage, cluster head (CH) selection is performed based on criteria such as node degree, neighborhood distance, remaining energy, and mobility, using the Weighted Clustering Algorithm (WCA) as a basis. These criteria are then optimized using the Gravitational Search Algorithm (GSA). In the second stage, node roles around the selected CHs are determined using the Enhanced Density-Based Spatial Clustering of Applications with Noise (Enhanced-DBSCAN) algorithm. This approach reduces parameter dependency and allows the formation of more flexible and balanced clusters. Simulation results show that DMHAKA significantly extends network lifetime and improves the packet delivery ratio. In comparative analyses with the WCA algorithm, the CH change rate in DMHAKA was found to be 40% lower, and the average CH lifetime was approximately 60% longer. Moreover, the average cluster lifetime results indicate that DMHAKA forms more stable clusters, directly contributing to energy savings and longer network life. These results reveal that DMHAKA provides a stable clustering structure that requires fewer reconfigurations. In addition, comparative analyses with the EE-WCA, E-MAVMMF, TSDR, and MORS-ASO algorithms demonstrated that DMHAKA exhibits superior performance in key metrics such as average remaining energy, end-to-end delay, packet delivery ratio, and throughput. The findings indicate that DMHAKA offers a scalable, stable, and energy-aware clustering approach when compared with various clustering algorithms. In this regard, DMHAKA can be considered a strong alternative for enabling efficient data transmission and achieving long-term sustainable network performance under dynamic MANET conditions.