Traffic Flow Prediction Model In Data Center Networks By Deep Learning Approach

A huge number of studies and researches on the prediction of data traffic on Data Center Networks(DCNs)focus on traffic engineering and usage of bandwidth.It is i mportant for accurate and timely traffic of information for communication purpose.W e have to do something reliable for many applications,such as identification of data,a void the collapse of data or handle the data just like congestion control,bandwidth spa ce and check list before the communication.From the last few years,flow of internet data have been blowing up and have moved in the time of big data for communication.We have to introduce prediction model for the data traffic flow.In this research work we have the comparative study about the prediction of data traffic.The proposed work is to check the architecture and feasibility of the Convolution al Neural Network with Long Short Term Memory(CNN-LSTM)and Gated Unit Rec urrent(GRU).These both prediction models have the ability to capture the data of spa tiotemporal and time dynamic which is most important factor for the communication t o pass all types of data traffic.In this scenario,firstly data is converted into Traffic M atrix(TM)form and then allow passing from the predicted models.During the passag e of data,the redundancy and weight of data is monitored.The data which is not in th e range or monitoring window according to the applied filter can be neglected.After passing through the layers in both predicted models CNN-LSTM and GRU,all the flo w is up to on the control of data sequences which decides the prediction on the basis o f previous and current flow value.In both cases,GRU has the good impact than the CNN-LSTM because the numb er of layers in GRU is less than CNN-LSTM and has optimization achievement also b etter than the CNN-LSTM due to less number of neurons inside the architecture.