Abstract: To improve the delivery rates, control overhead rates and reduce the average delay is the ongoing research in the random opportunistic networks moving scenes. Due to the opportunistic network structure is sparse and the network topology is variable, the efficiency of single copy routing forwarding strategy is very low. This paper defines a forwarding strategy based on Markov decision by combining the similarity between Brownian motion of pollen and nodes random movement in opportunity networks and analyzing the law of the random motion of nodes. In the case of an appropriately growing average delay, the strategy can control the overhead rates and advance the delivery rates. Finally, this paper verifies the correctness of the theoretical models by simulation experiments.

Key words: opportunistic network, Markov decision, delivery rate

摘要： 在机会网络节点随机移动的场景中，提高路由算法性能评价中的投递率，控制开销率，降低平均迟延是持续的研究方向。由于目前机会网络结构稀疏和拓扑多变，单副本路由转发策略效率较低。通过结合花粉布朗运动与机会网络节点的随机运动的相似性，并分析节点随机运动的规律，定义了一种基于马尔可夫决策过程的节点转发策略。该策略在平均延时适当增加的情况下，可以有效控制网络开销率，提高消息投递率。最后通过仿真实验验证了理论模型的正确性。

关键词: 机会网络, 马尔可夫决策, 投递率