A novel approach based on bio-inspired efficient clustering algorithm for large-scale heterogeneous wireless sensor networks

In large-scale heterogeneous wireless sensor networks (WSNs), clustering is particularly significant for lowering sensor nodes (SNs) energy consumption and creating algorithm more energy efficient. The selection of cluster heads (CHs) is a crucial task in the clustering method. In this paper, optimised K-means clustering algorithm and optimised K-means based modified intelligent CH selection based on BFOA for large-scale network (lar-OK-MICHB) is hybridised for CH selection process. Here, we utilised the extended capabilities of OK-MICHB algorithm for large-scale network. Furthermore, in many applications where energy is a primary constraint, such as military surveillance and natural disaster prediction, the stability region is also a significant factor, with a longer network lifespan being a primary requirement. In the proposed approach, only the CH selection is made after every round in place of cluster and CH change as done in conventional hierarchical algorithm. The simulation results reveal that, while keeping the distributive structure of WSNs, suggested lar-OKMIDEEC can locate real greater leftover energy nodes for selection of CH without utilising randomise or estimated procedures. Furthermore, as compared with the multi-level MIDEEC protocol, this offers a larger stability region with 68.96% increment, more consistent selection of CH in every round, and greater packets (i.e., in numbers) received at the base station (BS) with a longer network lifetime with 327% increment.