An improved clustering algorithm based on K-means++ and Knn method for underwater wireless sensor networks

The underwater wireless sensor network has become a new research hotspot in today's marine field, in which the clustering algorithm is a key problem in the underwater wireless sensor network. The LEACH algorithm, as a classical algorithm in the clustering algorithm, has the problems of uneven distribution of nodes due to the randomness of cluster head selection and the lack of consideration of node residual energy, which leads to the premature death of some nodes and affects the survival cycle of the network. In this regard, we propose the new algorithms that integrate the uniform distribution of cluster head nodes and the residual energy of sensor nodes - KK_LEACH algorithm and KK++_LEACH algorithm. KK_LEACH algorithm first uses the number of surviving nodes and the threshold value of LEACH algorithm to K-means++ clustering algorithm. The K-means++ clustering algorithm is used to find the k clustering centres by specifying the k value, and secondly, the m nodes with the highest remaining energy around them are selected as the first cluster nodes by the Knn (K nearest neighbour) algorithm. The KK++_LEACH algorithm considers the optimization strategy about the average energy of the network based on the KK_LEACH algorithm. KK_LEACH algorithm alleviates the imbalance of the remaining energy distribution of the sensor nodes compared to the LEACH algorithm and extends the network lifetime, while the KK++_LEACH algorithm further considers the optimization strategy about the average energy of the network based on the KK_LEACH algorithm to further extend the network lifetime and alleviate the energy distribution imbalance problem.