Research On Multi-objective Intelligent Optimization Design Software And Optimization Algorithm

As new services emerge,such as the Internet of vehicles,intelligent home,and virtual reality,network carrying capacity and transmission efficiency are faced with higher requirements,and the new technologies promote optical networks to advance toward ultralarge capacity and ultra-high speed.For network operators to provide network planning and optimization solutions,the performance of optical network resource optimization algorithms is critical,and researchers urgently need multi-objective intelligent optimization design software applicable to optical network scenarios to improve the reliability of network planning and optimization solutions.Network operators can achieve flexible expansion of network architecture and appropriate utilization of optical network resources in optical Dense Wavelength Division Multiplexing(DWDM)networks with the help of software defined networking technology,where the Multi-Controller Placement Problems(MCPP)are one of the current DWDM research hotspots.By dividing optical networks into multiple control domains and deploying a controller in each domain,the capital cost and operation cost of optical networks can be lowered to the utmost and the load variance of networks can be reduced maximumly.This thesis relies on the sub-project of the National Key Research and Development Program of China entitled "Broadband Communication and Novel Network",namely research on the application of novel 100G/200 G optical networks with engineering demonstration(2019YFB1803605).It designs and implements multi-objective intelligent optimization design software,and proposes multi-objective evolutionary algorithm with multi-strategy hybrid to solve MCPP.The first chapter of the thesis provides a brief introduction of optical network and optical network planning,introduces the development of optical networks,the architecture and components of DWDM network.The second chapter summarizes the design process and methodology,and describes the characteristics,functionalities and application scopes of mainstream network simulation software and multi-objective optimization software.It also presents the main optimization objectives of MCPP and the research status of mainstream solution algorithms.The main work and achievements of this thesis are as follows:(1)Design and implementation of multi-objective intelligent optimization design software.This thesis selects the appropriate network simulation software and multi-objective optimization software as a reference,condenses the primary requirements of the software and prioritizes each requirement.Secondly,it determines the software architecture,completes the functional design of its three modules,with include environment configuration,simulation operation and statistical output.It then chooses the appropriate programming languages and software development tools to implement specific programming for each functional module.Finally,the algorithms of solving multicontroller deployment problem in optical networks is embedded into the software.Multiple test cases have been designed to thoroughly test the software,and several flaws in the software have been improved based on the test results.(2)A multi-objective evolutionary algorithm with multiple hybrid strategies is proposed and studied.In this thesis,an MCPP model with optimization objectives that minimizes simultaneously the deployment cost,load variance and maximum propagation delay is constructed.It proposes a multi-objective evolutionary algorithm with multiple hybrid strategies to solve the model.The proposed algorithm include the following key features: a)using opposition-based learning to generate the initial population;b)using the adaptive crossover mutation operator to generate the offspring population;c)selecting individuals into the next generation by non-dominated sorting and two-stage measurement;d)using the comprehensive weight method to correct illegal individuals,and e)adding an adaptive catastrophe disturbance mechanism,the catastrophe cycle can be adjusted according to different evolutionary effects.Through the simulation of the proposed algorithm and the benchmark algorithm in Internet OS3 E,Interroute and Cogentco network topologies,it can be observed that the proposed algorithm achieves Pareto solution sets with better convergence,wider distribution and richer diversity,fully validating the effectiveness of the proposed algorithm.