Research On Service Discovery And Management Technology In Tactical Resilience Network

The tactical resilience network(TRN)is an important application scenario for future battlefield communication.With the continuous increase of network scale,the improvement of resource sharing requirements,and the proposal of man-machine cooperative combat mode,traditional connection-oriented communication networks have been difficult to support the tactical and resilience operational characteristics.The specific performance is that the communication capacity cannot keep up with the increase of network scale,and tactical features often impose harsh constraints on communication conditions.To this end,moving up from the physical layer,there have been gradual emergence of research and exploration in many dimensions such as storage and computing in theory,expanding the ability of intelligent information dissemination in the network.Moving down from the business layer,on the network structure,the design concept of data forwarding and control separation of Software Defined Network(SDN)and other structures has gradually emerged,and the network virtualization and interface openness capabilities have been gradually established,which comprehensively solves the development bottleneck problems such as expansion difficulties and insufficient flexibility in the network,and promotes the evolution of the network to a dynamic and flexible direction.Therefore,how to introduce intelligent and flexible working features in the future network to fully support the flexible characteristics of tactical networks will become an important research direction in the future.This paper introduces a tactical service-oriented architecture(TSOA)based on the characteristics of tactical resilience network.By combining the design concepts and methods of service-oriented architecture(SOA),the service architecture is requested on demand.Different from the connectionoriented communication structure,the above service structure can improve the information transmission efficiency,and overcome the lack of flexibility such as service delay and server pressure in the centralized control scheme under tactical resilient conditions.Specifically,under the TSOA service framework,this paper focuses on the decentralized service management scheme aiming at improving the efficiency of distributed service management,and studies the distributed intelligent service search method to improve the node's decision-making ability to reduce the pressure to connect the communication concept.This paper designs a new service management scheme based on partition strategy.Fully considering the node's centrality,service popularity,service size,service life duration,service availability and service forwarding probability,it divides the cache space of the node into two parts: the service cache area and the relay buffer area.Combing the different characteristics of service stored by the two areas,different service caching strategies and service replacement strategies are designed to achieve more intelligent service management.This paper implements the design scheme in ONE,and compares the performance with the service management scheme's in the traditional tactical communication network,which verifies the superiority of the design scheme.This paper makes full use of the storage capacity and computing power of network nodes,and proposes a service search algorithm based on learning process.Utilizing the Random Forest algorithm combined with active learning,the node has the ability to make decisions and can make optimal decisions based on the state of the real-time system.Using the decision model obtained through the training,the node can select the most suitable data packet for forwarding and storage,thus improving the forwarding efficiency of data packets and promoting the service search efficiency of the system.In addition,the node sends the nodes encounter information and updates the network information when a forwarding event occurs,so that the system has higher adaptive characteristics.In this paper,the service search algorithm based on the Random Forest algorithm is implemented by ONE simulator,and the simulation experiment is compared with the existing service search algorithm.The service search method designed in this paper is proved to be efficient and adaptive.In summary,this paper studies the problem of service discovery and management under tactical resilience network,and proposes the idea of using distributed non-registered architecture to build a tactical serviceoriented architecture.This paper focuses on the use of learning processbased methods to improve service management efficiency and service selection efficiency.