| With the development of Internet of things,Narrow Band Internet of Things(NB-Io T)has become a research hotspot in recent years because of its advantages of low power consumption,wide connection and other characteristics.However,NB-Io T also has the disadvantage of low throughput.The research found that the packet data convergence layer in NB-Io T mainly implements header compression function of IP data stream.However,the current header compression algorithm can not be well adapted to the complex and changeable channel environment of NB-Io T.Therefore,it is necessary to improve the header compression function of this layer to increase the header compression rate,so as to improve the throughput of the system.Therefore,a header compression algorithm named ROHC-DBN based on dynamic Bayesian network prediction model is proposed.The core idea is to establish a dynamic Bayesian network prediction model by studying the state transition process of the decompression end.According to the model,combined with the data packet transmission status and the state transition probability table,the decompression state can be predicted.The algorithm can keep the state synchronization better between the compression end and the decompression end,and improve the header compression rate.Afterwards,based on the improved algorithm ROHC-DBN,the entire packet data aggregation layer is implemented and the implementation of the header compression function module is emphasized.In this algorithm,the state transition of header compression will not depend on the principle of optimistic approach and the principle of periodic regression any more,but obtain the feedback information of mixed automatic retransmission request through the interface with MAC layer to get the transmission status of packets instead,and then state estimation can be realized according to the dynamic Bayesian network prediction model.In this thesis,the state transition probability is obtained by simulation,and then the validity of the algorithm is verified by actual measurement.By compiling the protocol code of the packet data aggregation layer and building ARM hardware test platform,the average compression header length and the throughput are tested respectively.The test results show that compared with the original ROHC algorithm,the average compression header length is reduced by more than 0.2 bytes,the throughput is improved by 5.01%.Therefore,it can be concluded that the ROHC-DBN algorithm does optimize the performance of the system and has a certain value for the application expansion of NB-IOT. |
