Traffic Characteristic Analysis And Dynamic Routing Algorithm Research Of Intelligent Optical Network

The vigorous development of broadband services has enriched the content supply of information services.At the same time,user scale and business traffic entered an explosive growth stage.As a transmission tool for carrying all kinds of data in the information society,the optical fiber communication network directly affects the user’s Internet experience.Therefore,it is of great significance to reasonably and effectively perform performance optimization based on the existing optical communication network architecture.Using inherent experience to design a network construction scheme fails to reflect the macro characteristics of user behavior and service flow,and is less instructive for tracing traffic distribution and perceiving user characteristics.At the same time,the network demand has long-term dynamics with the changes of operating indicators such as user penetration rate and service growth rate,and the static planning method lacks foresight,which may cause the phenomenon of resource disconnection.In addition,service accumulation is prone to occur in the process of data exchange.How to flexibly and dynamically distribute burst services evenly to each link of the entire network has become an important network optimization goal.In view of the above problems,this paper presents the planning scheme and optimization strategy of intelligent all-optical network from three aspects:network infrastructure adaptability,bearing capacity estimation accuracy,and bandwidth resource allocation flexibility.First of all,combined with the massive existing network data collected by operators,this paper uses statistical analysis,k-means,XGboost and other mining methods to extract user characteristics,and strives to obtain "regional labels" of traffic distribution and target users with high traffic characteristics,so as to assist with network governance.Combined with the above analysis indicators,the paper adopts simulated annealing algorithm and PON migration algorithm based on user space distribution and traffic characteristics for ODN and PON network structures respectively.The construction planning of the optical access network and the rational dredging of business traffic in space have been completed,and the resource utilization rate and broadband service capability of the underlying network have been improved.Then,combined with the user package type,this paper presents a traffic estimation method under time series.By analyzing and calculating the qualitative and quantitative relationship between the cross-sectional traffic of PON equipment and the number of users of each type of packages,a passive optical network traffic estimation method based on the Prophet model is proposed.The error of a single PON port is maintained below 2%.It not only provides the possibility for pre-deployment of equipment,improves the bearing stability of the optical access network,but also lays a foundation for subsequent traffic grooming.Further,this paper focuses on the resource allocation strategy of the upper-layer service bearing core network,and provides the traffic planning methods of the optical network at different levels.This paper combines the SDN bearer architecture of the optical core network,makes full use of the advantages of the separation of the control plane and the data plane of the optical core network,and introduces the reinforcement learning theory into the routing strategy of the network element controller.The experimental results show that this method effectively alleviates the problem of sudden service accumulation,better achieves the goal of load balancing,and further improves the matching degree of optical network resources and broadband service requirements from the software level.The algorithm provides a new idea for intelligent operation and maintenance and precise planning of optical networks.