Abstract: Aiming at the problem that the significant disparities in  residual energy among nodes led to uneven energy consumption,  affecting  load balancing, limiting  information transmission efficiency, and  increasing packet loss rates  in wireless sensor networks employing non-uniform clustering, we  proposed a load-balanced non-uniform clustering routing algorithm for wireless sensor networks. Firstly, we used a hybrid energy-efficient distributed clustering protocol to perform non-uniform clustering of nodes in wireless sensor networks,  calculated the clustering probability for each node and selected the node with the highest probability as a candidate cluster head node. Secondly, we designed a node  evaluation function based on  residual energy of nodes and its distance to neighbors, calculated the evaluation function value of each node, elected the  node with the highest evaluation function value as the  cluster head  to construct a cost function,  selected relay nodes to enhance transmission efficiency, and  formed an inter-cluster multi-hop routing algorithm. Finally, we distributed the data transmission tasks to  multiple nodes in the network to  reduce single-node energy consumption and achieve  non-uniform clustering routing load balancing in wireless sensor networks. Experimental results show that  the residual energy of each node in this algorithm  only  fluctuates around 6 J, and the fluctuation amplitude of the curve is small. The network packet loss rate is low at 0.22% when   transmitting 2 000 bit data volumes, indicating high  load balancing.

Key words: load balancing, HEED protocol, non-uniform clustering, node evaluation function, wireless sensor network