Research On The Key Technologies Of The MAC Layer Of Radio-over-Fiber Industrial Interconnection Network

With the rapid development of access network technology in the new era,in the trend of broadband and wireless,the Radio-over-Fiber network comes into being,it can not only transmit data at high speed,but also improve the coverage of wireless communication network,and has the advantage of easy dynamic control and maintenance.It's inevitable there has a huge application prospects In the future industrial Interconnection control network.The key technology research on the MAC layer of the Radio-over-Fiber network and aiming at the high speed transmission characteristics of the Radio-over-Fiber network,which can match the MAC layer channel access technology,is believed to be of great application value in the future industrial Internet.In this paper,the efficiency of the existing MAC channel Access technology algorithm and the adaptability to the complex environment is researched and implemented,two improved methods are proposed for the different number of nodes in the network environment,and the simulation experiments are carried out using MATLAB.The specific research work is as follows:Firstly,a dynamic P-persistent CSMA improved algorithm based on slip statistics is proposed to solve the disadvantage that P-persistent algorithm probability cannot change with environment.P-persistent algorithm is widely used in node random competitive access channel,and how to choose the probability of sending p to maximize protocol performance is the key problem of P-persistent CSMA.It is found that using fixed p value is difficult to guarantee the protocol performance under different network load conditions,so it can not be applied in the high-speed transmission of the Radio-over-Fiber network.In order to solve this problem,a dynamic P-persistent CSMA algorithm based on slip statistics is proposed,which adopts the recursive slip statistic method to realize the dynamic adjustment of sending probability.From the simulation experiment,we can see the better effect,the efficiency is obviously better than the other three kinds of CSMA algorithms.Secondly,the node state space complexity of the Q-Learning algorithm increases exponentially with the increase of the number of nodes and the Q-Learning algorithm has the phenomenon of repetitive learning and low learning efficiency.This paper proposes a CSMA optimization strategy based on shared experience in the process of multistate Q learning.How to reduce the number of dimensions and reduce the amount of search and computation is the key point when the node has limited capacity.At the same time,Q learning is an unsupervised online learning,prior knowledge of the environment is not required,but also because such nodes need to spend a certain amount of learning time to accumulate experience,and for the node,whether it is action,or state,or even rewards are independent,can not adapt to the complex environment in time.To solve these problems,this paper proposes a CSMA optimization strategy based on the shared experience in the multistate Q learning process.In this optimization strategy,the state space is defined for the node,and the node decision-making process is simplified,while the nodes learn by themselves and share the experience periodically.Experiments show that the optimization strategy reduces the complexity of the state space,reduces the search volume,shares experience in the learning process,and utilizes the experience knowledge better.