Study On Forward Error Correction And Sevices Identification Of 10G EPON Protocol Analyzer

With the increasing demand for network bandwidth,10 G EPON technology has been developed rapidly in recent years and has been deployed in some pilot area while the research of 10 G EPON protocol analyzer is lagging behind.The conventional 10 G EPON protocol analyzer just decodes data from the physical layer of PON system,and then packages and outputs the corresponding ethernet protocol packet.This single function of 10 G EPON protocol analyzer leads to a consequence that the conventional 10 G EPON protocol analyzer becomes just an equipment which is used to capture ethernet protocol packet.When it comes to know the channel condition of network and services carrying and achieve the fine control and management of a variety of protocols and services,the analysis of upper layer protocols and services is fairly necessary.In consideration of the defects of conventional 10 G EPON protocol analyzer,this thesis proposes a new 10 G EPON protocol analyzer scheme.The new scheme includes two parts: Ethernet protocol analysis and services identification.The former part captures 10 G EPON protocol packet and its most complex module is to achieve forward error correction decoding.The latter part aims to classify different services through the relevant technology,which could understand the network traffic distributions of PON access network and provide the foundation for the realization of intelligent pipeline and improvement of QoS.In the light of the two key technologies of the new 10 G EPON protocol analyzer,the research work of this thesis is divided into the following two parts.(1)According to the principle of forward error correction decoding,QuartusII13.0 and Verilog are used to design and simulate the circuit code and the RS(255,223)decoding circuit designed and achived by this thesis could meet the performance requirements of 10 G EPON.The resource consumption of decoder circuit is attractive and the overall delay is only 72 cycles while maximum error correcting capability of the decoder is 16 error codes.(2)This thesis does some investigations and contrasts on the services identification technology.In view of the defects of service identification technology in existing PON network,this thesis proposes a novel scheme based on deep learning technology.In order to verify the scheme,this thesis uses CNN and DAE to achieve services identification and improves the CNN algorithm and the training process of these two algorithms based on the actual situation.Finally,this thesis uses MATLAB to adjust network structure and complete the simulation experiment.The results show that the accuracy rate of the two algorithms both can be more than 97% and 99% in public data sets and private data sets respectively,but DAE has an advantage over CNN in training time.The simulation experiment results of this thesis also demonstrate that CNN and DAE have excellent classification ability for feature data and own an outstanding ability to extract features for services data.Therefore,the services identification scheme based on deep learning for 10 G EPON protocol analyzer proposed by this thesis has an advantageous application value.