Dynamic Optimization of Sphincs+ Signature Algorithm Parameters Based on A Ucb-Like Algorithm

This study addresses the issues of redundant security and suboptimal resource efficiency in fixed parameter configurations of the post-quantum cryptographic algorithm Stateless PHoton-based Isogeny Nihil Crux Signature+ (SPHINCS+). Observing that the key generation overhead in this signature scheme is significantly lower than that of signature creation and verification processes, this work proposes a dynamic parameter optimization method based on a UCB-like algorithm. By modeling core parameters, including the total height of the HyperTree, the number of secret keys in Forest of Random Subsets (FORS) trees, and the leaf count of FORS trees as adjustable variables, our approach implements intelligent parameter selection through a multi-armed bandit framework. The algorithm autonomously balances cryptographic strength, signing speed, and storage costs by learning user signing patterns while adapting to real-time security requirements and system resource availability, thereby eliminating computational waste and excessive security design inherent in static parameter configurations. Experimental results demonstrate that this method achieves a 6% to 9% reduction in signature generation resource consumption while maintaining scenario-specific security guarantees. This research provides new optimization perspectives for enhancing the practical engineering applicability of the SPHINCS+ hash-based signature scheme.