The　Octagonal　Steiner　Minimal　Tree　(OSMT)　problem　is　an　NP-hard　problem,　which　is　one　of　the　key　problems　in　non-Manhattan　routing.　Particle　Swarm　Optimization　(PSO)　has　been　proved　to　be　an　efficient　intelligent　algorithm　for　optimization　designs.　This　paper　presents　an　OSMT　algorithm　based　on　discrete　PSO　(DPSO),　namely　OSMT_DPSO,　to　optimize　the　wire　length.　In　order　to　solve　the　problem　of　the　slow　convergence　rate　of　PSO　used　for　a　high-dimensional　space　optimization,　a　self-adapting　strategy　that　can　adjust　the　learning　factors,　and　combine　with　the　crossover　and　mutation　operators　of　Genetic　Algorithm　(GA)　is　proposed.　The　experimental　results　show　that　the　proposed　algorithm　can　efficiently　provide　the　solution　of　OSMT　problem　with　good　quality.　Moreover,　the　algorithm　can　obtain　several　topologies　of　OSMTs　which　is　beneficial　for　optimizing　congestion　in　VLSI　global　routing　stage.