Markov Finite Horizon Decision Algorithm of Shortest Path Tree
最短路径树的马尔可夫有限阶段决策算法

The idea of decision-making is focused on in this paper.Combining with Markov decision process theory,a SPT finite horizon decision model is established.A new auxiliary graph:Reverse Graph is introduced,combining with which the theoretical solving algorithm of the model is modified,a SPT reverse recursion iterative algorithms is proposed and its validity is proved.Base on these,an improved model and algorithm without Reverse Graph are also put forward,and time and space complexity of the two algorithms are analyzed.The theoretical analysis results show that these algorithms are characteristic by distributed parallel computation,which can balance workload between all nodes,reduce time and space complexity,and is loop-free.So it can be effectively applied in the embedded connection environment with limited resources and the peer-to-peer network environment.