Optimized Rosenblatt (Hermite) process using CNN-LSTM neural network for NoC communication
The role of the traffic model becomes important in order to understand and decipher the performance of networking issues. Various research models and synthetic based traffic models are designed to understand the issues and requirements of network to achieve Quality-of-Service (minimize latency, increased throughput etc.). Networks-on-chip (NoC) are becoming the solution for existing scalable multicore Systems-on-chip (SoC) for higher order communication management. There is no existing technique available for the generation of traffic models which captures the temporal and spatial deviation for the realistic traffic. In this paper we are proposing the deep learning method applied on the synthetic traffic models to optimize the spatial and temporal variations from the realistic traffic. Our models will provide the optimized traffic models which will have the same statistical properties as for the realistic models.
Keywords: CNN, LSTM, Traffic Modelling, Networks-on-Chip.