Dcaro: Dynamic cluster formation and AUV-aided routing optimization for energy-efficient UASNs

In Underwater Acoustic Sensor Networks (UASNs), optimizing energy efficiency and minimizing void occurrences in routing is paramount. Due to the energy constraints of sensor nodes, low-power transmission is essential for conserving energy. Previous research highlighted the effectiveness of clustering and routing to enhance energy efficacy in UASNs. Therefore, the clustering and routing processes can be considered as optimization problems that are nondeterministic polynomial-time (NP) hard. These challenges can be tackled through the application of machine learning algorithms and meta-heuristics. In this context, K-means clustering is employed to partition the network into clusters, designating the centroid as an ideal Cluster Head (CH) location. This ensures a one-hop proximity between the CH and cluster members, reducing transmitting power and enhancing network energy efficiency. Subsequently, a potential CH is selected using a marine predator optimization (MPA) algorithm based on the derived multi-objective fitness function. The MPA algorithm not only determines the optimal CH but also moves the elected CH to the K-means centroid location. Consequently, Autonomous Underwater Vehicles (AUVs) are utilized to collect and route packets from the CH to the Base Station (BS), minimizing the occurrence of void nodes and avoiding obstacle collisions. An optimal routing path for AUV is established through a way-point-based navigation scheme to achieve high packet reliability. Additionally, the proposed method (DCARo) dynamically determines the optimal number of clusters using the elbow method, ensuring scalability according to network size. Extensive simulations affirm the superiority of the DCARo across various performance metrics.