摘要
随着人工智能、网络通信、网络应用的飞速发展,传统的路由算法如RIP、OSPF在面对网络流量的指数级增长以及不同服务需求的情况下存在收敛慢、平均时延高等一系列问题。而近年来深度强化学习在复杂控制领域取得巨大发展。路由优化算法本质上讲就是一个控制优化问题。所以为了克服现有路由算法在某些场合下的弊端,将深度强化学习与计算机路由优化相结合。论文提出一种利用改进的DDPG算法,并命名为TD3OR算法去解决传统路由在某些场合下的弊端。实验表明,采用TD3OR算法的路由对比单纯DDPG算法以及传统OSPF算法的路由具有更低的延时,证明其是有效的。
With the rapid development of artificial intelligence,network communication,network application,traditional routing algorithms such as RIP and OSPF have a series of problems such as slow convergence and high average delay in the face of expo-nential growth of network traffic and different service requirements.In recent years,deep reinforcement learning has made great progress in the field of complex control.Routing optimization algorithm is essentially a control optimization problem.In order to over-come the drawbacks of existing routing algorithms in some situations,and combine deep reinforcement learning with computer routing optimization,this paper proposes an improved DDPG algorithm named TD3OR algorithm to solve the drawbacks of traditional routing in some situations.The experiment shows that the route using TD3OR algorithm has lower delay than that using DDPG algo-rithm and traditional OSPF algorithm,which proves that TD3OR algorithm is effective.
作者
郑艺
韩永国
ZHENG Yi;HAN Yongguo(School of Computer and Software,Chengdu Neusoft University,Chengdu 611844)
出处
《计算机与数字工程》
2025年第8期2117-2121,共5页
Computer & Digital Engineering
