Proceedings of the International scientific and practical conference ―New York Global Science Conference 2026‖ (March 6-8, 2026) / Publisher website: www.naukainfo.com. – New York, USA, 2026. - 250 p.

74 The set of nodes is divided into a set of primary clusters, where a first-level cluster head responsible for local traffic aggregation is selected. The selection of this node is carried out on the basis of an integral criterion that takes into account the residual energy of the node and its spatial position relative to other elements of the cluster. At the second level of the hierarchy, a network of inter-cluster interaction is formed. Cluster heads of the first level are combined into secondary clusters, for which second-level aggregator nodes are again determined. Traffic aggregation in a multilevel structure allows for a significant reduction in the number of inter-cluster transmissions. The effectiveness of the proposed approach was confirmed through computer simulation and comparative analysis with the baseline LEACH algorithm. The first stage of the experiments was aimed at analyzing the dynamics of the residual energy of the nodes over time. In the case of multilevel organization, a slower decrease in the average energy level is observed, which indicates a decrease in specific costs for routing and traffic relaying. The second stage of the simulation was devoted to assessing the uniformity of the energy load among the network nodes by analyzing the variance of the residual energy. The decrease in variance in the case of multilevel clustering confirms a more balanced distribution of roles among drones and the absence of rapid depletion of individual aggregator nodes. The key performance indicator of energy-saving algorithms is the network lifetime. An analysis of the number of active nodes over time shows that the moment of network degradation in the case of the proposed approach occurs later, which directly indicates an extension of its autonomous operation time. Summarizing the experimental results, it was found that with an increase in the network load coefficient , the average network lifetime decreases for both approaches. At the same time, the proposed multilevel cluster model in the entire studied range of from 0.2 to 0.9 provides an increase in network lifetime by approximately 15% compared to the LEACH algorithm.