With the development of UAV (unmanned aerial vehicle) networks toward clustering and expansive mission coverage areas, the contradiction between the increasing scale of communication and limited spectrum resources has become increasingly prominent. To enhance network robustness and spectrum utilization, the available spectrum resources are typically divided into multiple channels. However, in scenarios characterized by complex spectral environments or malicious interference, the actual available channels for geographically distributed UAV nodes exhibit significant heterogeneity. This disparity in channel availability poses significant challenges to the clustering process of UAV networks. Traditional single-channel clustering algorithms demonstrate limited applicability in such scenarios, while existing multi-channel clustering algorithms often rely on idealized assumptions of uniform channel availability across all nodes, thereby restricting their effectiveness in addressing practical networking challenges. To resolve the network formation difficulties caused by heterogeneous channel availability among UAV nodes, a multi-channel clustering algorithm based on adaptive node degree difference (MC_ANDD) was proposed for UAV ad hoc networks, which is designed to achieve efficient network organization and resource management through effective clustering of large-scale UAV nodes.

This study focused on the initial stage where UAV nodes maintained relatively stable geographical distribution, proposing an adaptive strategy for forming a clustered network topology. The proposed clustering algorithm MC_ANDD was developed based on a hierarchical clustering framework with modularity optimization, which innovatively improved node similarity computation under multi-channel conditions. This advancement effectively mitigated the networking constraints caused by heterogeneous channel availability among UAV nodes. Specifically, by incorporating an adaptive node degree difference into the computation of the Jaccard similarity coefficient, cluster formation was guided not only by the proportion of common neighbors between nodes, but also by their overall connectivity within the network. The algorithm further combined node similarity with modularity optimization to construct a network topology that better reflected the spatial distribution characteristics of UAV nodes. This algorithm effectively avoided the formation of excessively large or small clusters, which could otherwise result in severe resource contention or underutilization. The algorithm further ensured two critical constraints: each cluster shared at least one common communication channel to maintain intra-cluster connectivity, and each node was assigned to exactly one cluster.

The simulation results validate the effectiveness of the proposed algorithm under varying numbers of nodes and channels. Compared with the Fast Unfolding, Jaccard similarity based clustering network construction (JS_CNC), and hierarchical virtual clustering based multi-channel network construction (HVC_MCNC) algorithms, MC_ANDD constructs a more balanced clustering topology. In particular, it achieves a significant reduction in cluster size deviation under fixed channel numbers and varying node counts. In addition, under varying node and channel settings, the MC_ANDD algorithm demonstrates a significant advantage in improving network throughput, further validating its adaptability in multi-channel environments.

Based on the modularity-optimized hierarchical clustering algorithm, this study incorporated an adaptive node degree difference tailored for multi-channel environments into the computation of node similarity. The network modularity function was maximized to cluster large-scale UAV nodes, and network throughput was analyzed using the Bianchi model. Simulation results demonstrate that the proposed multi-channel clustering algorithm based on adaptive node degree difference for UAV ad hoc networks consistently achieves lower cluster size deviation and improves overall network throughput under varying node and channel numbers. A relatively balanced number of members within each cluster not only facilitates cluster head management but also enhances the efficiency of communication resource reuse. Such optimized network topology proves particularly valuable for mission-critical operations including intelligence gathering, reconnaissance, and search-and-rescue tasks, demonstrating significant application potential in multi-channel communication environments.