Research On Deep Reinforcement Learning Based Routing Algorithm For Wireless Multihop Networks

In recent years,wireless multihop networks have attracted much attention due to their wide perspectives in both military and civilian applications.Plenty of opportunistic routing algorithms have been proposed for the wireless networks during last decades.It is generally accepted that deep reinforcement learning will be a promising paradigm to enhance the performance of wireless networks.However,existing research on wireless networks and deep reinforcement learning is mostly conducted independently,which fails to exploit the learning capabilities of reinforcement learning techniques to optimize the performance of network routing adaptively.This has greatly limited the potential of deep reinforcement learning to improve the performance of wireless networks.To take advantage of the learning capability of deep reinforcement learning and also improve the intelligence of wireless networks while meeting the needs of fast development of wireless networks,this thesis propose a deep reinforcement learning based energy efficient opportunistic routing algorithm for the wireless multihop networks,which enables a learning agent to train and learn the routing policy to reduce the transmission time while balancing the energy consumption to extend the life of the network in an opportunistic way.Furthermore,the proposed algorithm can significantly alleviate the cold start problem and achieve better initial performance.Simulation results demonstrate that the algorithm in this thesis yields higher performance in terms of routing and energy consumption as compared with existing solutions for wireless multihop networks environment.The major work and contributions of this thesis are listed as follows:· Formulate the opportunistic routing problem in wireless multihop networks as an Markov Decision Process and define the corresponding state space,action space,and reward function.· Propose a novel energy efficient opportunistic routing algorithm using deep reinforcement learning to solve packet routing problem in wireless multihop networks,which can balance the routing performance and energy consumption effectively.· The proposed algorithm can effectively alleviate the cold start problem in the traditional DQN(Deep Q Network)based algorithm and can improve the performance at the early stage of the learning.· Extensive simulations are conducted and the results show that the proposed algorithm demonstrates appreciable gain as compared with existing work.