关键词: Steiner最小树, X结构, 绕障, 离散麻雀搜索优化, 超大规模集成电路

Abstract: The Steiner minimum tree is the optimal connection model for solving the routing problem of Very Large-Scale Integration. However, modern chips often contain various obstacles, such as macro cells and IP blocks, making the construction of Steiner minimum trees more challenging. Moreover, considering the excellent wirelength optimization capabilities of X-structure routing and the promising application of the sparrow search algorithm in solving NP-hard problems, this paper proposes a discrete sparrow search optimization-based X-architecture obstacle-avoiding Steiner minimal tree algorithm. First, a sparrow representation method based on edge-point pairs and an efficient fitness calculation method are proposed, coupled with a sparrow population update mechanism based on discrete mutation and crossover operations, which address the discrete X-structure obstacle-avoiding Steiner minimum tree problem. Second, a preprocessing strategy is proposed to avoid redundant calculations. Third, a hybrid initialization strategy is introduced, which combines the greedy approach and the roulette wheel selection method to enhance the diversity of the initial population. Fourth, an adjustment strategy based on detours is developed to meet obstacle constraints. Finally, a mixed refinement strategy is proposed to further minimize the wirelength cost. Experimental results show that the proposed algorithm achieves superior wirelength optimization compared to similar works.

Key words: Steiner minimum tree, X-architecture, obstacle-avoiding, discrete sparrow search optimization, VLSI