| With the rapid development of rail transit,the traditional Train Communication Network(TCN),due to its limited bandwidth,has been unable to meet the growing demand of both train control and passenger services simultaneously.Therefore,it is urgent to develop the Train Control and Service Network(TCSN)system.The aim of TCSN is to establish a system with large bandwidth,high reliability,high reconfigurability and high intelligence,which can integrate train control and passenger service business.The article relies on the National Science and Technology Support Program(2015BAG14B00)for related research,which references the basis of the traditional computer network and combines with the characteristics of the TCSN system to improve and simulate algorithms.Firstly,the characteristics of each service in the TCSN system are summarized and classified.Based on the traditional computer network methods for traffic simulation,the self-similar Fractal Brown Motion(FBM)model is used for the TCSN system.And the rationality of the FBM model is explained.Secondly,based on the analysis of two common traffic shaping control algorithms and the characteristics of the TCSN system,the token bucket algorithm is improved.The overflowed token bucket is added,which is used to store the overflowed tokens.The current packet loss rate of each bucket,the number of cached packets and the number of tokens in the bucket are taken as evaluation indicators,the entropy weight method is used to calculate the weight by these indicators,and the weighted max-min fair algorithm is used to distribute tokens of the overflowed bucket by the weight calculated before.Last through the comparison of simulation,the result indicates that the improved token bucket algorithm can not only reduce the packet loss rate but also reduce the waste of token resources.Finally,the scheduling algorithm in the network is studied,and the Weighted Fair Queueing(WFQ)algorithm is focused.Combined with the characteristics of the TCSN system,a dynamic weighted fair queueing algorithm based on the queue length(DQLWFQ)is proposed,which adds a classifier and queue length recorder in the WFQ.The classifier can distribute the data packets to their respective queue,and the recorder can timely feed back the real-time length of each queue to the scheduler.Then the scheduler can dynamically distribute the bandwidth based on a comprehensive evaluation of the queue length and priority to ensure the transmission of all kinds of business.Last,the WFQ algorithm and DQLWFQ algorithm are compared and simulated under the normal traffic situation and the burst traffic situation.The results show that the DQLWQ algorithm can not only distinguish the business but also can improve network parameters of the real-time business and the higher priority business,such as delay,packet loss rate,and throughput,etc.,so the simulation results have engineering application value for the establishment of TCSN network. |
