Performance of Synthetic Rosenblatt Process under Multicore Architecture

The Networks-on-Chip provides the solution for communication amongst different processing units placed on a single chip through the interconnection network for Multiprocessor Chip or Multiprocessor System-on-Chip (MPSoC). However, the simulation of real-time traffic is complex and requires a lot of machine-hours. So the need for synthetic traffic becomes crucial for its flexibility and fast simulation time which saves a lot of machine-hours. In this paper, we proposed the use of non-Gaussian Rosenblatt process for generation of synthetic traffic as compared to Gaussian process. We have analyzed the performance of the Rosenblatt process designed from different Hurst parameter index for different traffic patterns. Our study provides help to optimize & selection of network resource units used for communication for the network designer.