Research And Optimization Of QoS Mechanism In HINOC System

HINOC technology is an access network solution designed for coaxial networks with fully independent intellectual property rights.The transmission rate is up to 1Gbps.However,with the evolution of technology and the development of multimedia services,the integration of various network systems and the hybrid networking of different technical frameworks are facing the requirements and challenges of digitization,networking,high bandwidth and low latency.In the case of complex data services,the gigabit access rate of HINOC2.0 has become difficult to guarantee the quality of service for users.In order to improve the service quality of the coaxial system,this thesis researches and optimizes the Qo S mechanism of the HINOC2.0 system,which provides certain reference value for the later development of the Qo S mechanism in the HINOC3.0 standard.Aiming at the above problems,this thesis investigates the Qo S mechanism of the HINOC2.0system and several classical scheduling algorithms.Based on the service characteristics of the HINOC2.0 system,this thesis proposes a changeable weight polling scheduling algorithm based on the real-time size of queues(PCDWRR).The algorithm adds a queue manager and a weight regulator on the basis of the DWRR algorithm.The manager feeds back the real-time size of each queue to the regulator in time,and the regulator dynamically adjusts the weight according to the situation of the queue,ensuring the reasonable allocation of bandwidth resources.At the same time,aiming at the phenomenon of token overflow waste in the traditional token bucket algorithm,a shared token algorithm suitable for the HINOC2.0 system is designed by combining the PCDWRR algorithm with the token bucket algorithm.The algorithm adds overflow bucket,token manager and token allocator on the basis of token bucket algorithm.The overflow bucket collects the overflowed tokens,and the manager counts the number of assignable tokens and feeds them back to the allocator,who adjusts the token allocation rules in real-time according to the PCDWRR algorithm to ensure the reasonable utilization of token resources.In order to verify the effectiveness of the two improved algorithms,this thesis builds a HINOC2.0single-channel model based on the OPNET simulation platform,and the queue scheduling module and traffic policing module are designed in the model.The DWRR and PCDWRR algorithms are implemented based on scheduling module programming,token bucket algorithm and shared token algorithm are implemented based on traffic policing module,and the improved algorithm is comprehensively simulated and analyzed from a variety of service traffic scenarios.The simulation results show that compared with DWRR algorithm,PCDWRR algorithm optimizes scheduling logic,improves latency and packet loss rate of important services,and improves bandwidth utilization by11.8%.The shared token algorithm improves the waste of token resources in traditional token bucket calculation,improves token utilization by 14.9%,and significantly reduces the packet loss rate.In order to further verify the actual performance of the improved algorithm,this thesis carries out functional simulation and board-level verification for the two improved algorithms based on the HINOC2.0 hardware verification platform,that is,to verify the correctness of the improved algorithm’s functional logic from two aspects of sequence waveform and flow testing.The verification results show that the two improved algorithms designed in this thesis significantly enhance the quality of service of the HINOC2.0 single-channel system,and provide certain theoretical guidance and technical support for the subsequent research and development of HINOC3.0 technology.