Research On Routing Mechanism Of Software-defined Internet Of Underwater Things

The huge demand for marine exploitation has led to the thriving of the Internet of Underwater Things(IoUT)and attracted continuous attention from the research community.This growing interest can be largely attributed to new civilian and military applications brought about by large-scale underwater equipment,such as underwater static sensors,Autonomous Underwater Vehicles(AUVs),and Remotely Operated Vehicles(ROVs).The current IoUT is essentially hardware-based,depends on closed and rigid architecture design,and lacks scalability and flexibility.This brings great challenges to underwater routing and hinders the provision of truly differentiated services to highly diversified underwater applications.Software-defined Networking(SDN)is recognized as the next-generation network model,which relies on a highly flexible,programmable,and virtualized network architecture to significantly improve network resource utilization and simplify network management.This paper introduces the "SDN+AI" paradigm into the routing research of IoUT,establishes the architecture of multi-controller software-defined IoUT,and proposes a Quality-ofService(QoS)routing algorithm based on reinforcement learning.Finally,in-depth simulation experiments are carried out to verify the effectiveness of the proposed scheme.The main research work and innovative achievements of this paper are summarized as follows:(1)In order to solve the problem of reliability and performance bottleneck of singlecontroller software-defined network,a multi-controller software-defined IoUT architecture is designed.Firstly,the multi-controller control plane for IoUT is studied,a control domain division method based on improved BIRCH algorithm is proposed,and the relevant SDN entity underwater communication solution is designed.Those measures improves the network throughput and prolongs the network life.Furthermore,a load balancing algorithm based on switch migration is proposed,which aims to effectively allocate switches to underutilized controllers.This paper uses a multi-criteria decision-making method,that is,Technique for Order Preference by Similarity to Ideal Solution(TOPSIS)to select the target controller,which realizes the reasonable distribution of network load and the balance among communication overhead,response time and computational complexity.(2)In order to solve the problem of various QoS requirements of different types of underwater applications,this paper adds a knowledge plane to the above architecture and proposes a QoS-oriented adaptive routing algorithm based on deep reinforcement learning.Firstly,the deep reinforcement learning network model,reward function design and security learning strategy based on Deep Deterministic Policy Gradient(DDPG)framework are studied.Furthermore,the methods of network information collection,rule distribution and flow table installation are proposed,which are suitable for multi-controller plane.The reinforcement learning agent can use the global view to consider the link state information to make QoS routing decisions.The results show that the proposed routing algorithm has a great improvement in convergence rate,packet loss,delay,energy consumption,and achieves a QoS satisfaction rate as 95%.