An　improved　particle　swarm　optimization　(PSO)　algorithm　for　data　streams　scheduling　on　heterogeneous　cluster　is　proposed　in　this　paper,　which　adopts　transgenic　operator　based　on　gene　theory　and　correspondent　good　gene　fragments　depend　on　special　problem　to　improve　algorithm＇s　ability　of　local　solution.　Furthermore,　mutation　operator　of　genetic　algorithm　is　introduced　to　improve　algorithm＇s　ability　of　global　exploration.　Simulation　tests　show　that　the　new　algorithm　can　well　balance　local　solution　and　global　exploration　and　is　more　efficient　in　the　data　streams　scheduling.　©　Springer-Verlag　Berlin　Heidelberg　2007.