Dynamic Scheduling of Real-time Multi-core Subtask Based on Traffic Prediction

Subtask scheduling on multi-core is an important direction in the world today. There is little research of dynamic scheduling subtask according to the network traffic prediction. This paper proposes a method of dynamic scheduling of real-time multi-core subtask based on network traffic prediction. According to the prediction result of network traffics, the larger tasks scheduled dynamically into some subtasks. In the simulation, we scheduled the subtasks in the multi-core network software pipeline architecture, according to the prediction of Markov Model of media data streams. Eventually reach in the case of packet drop rate is acceptable, and CPU utilization rate has fallen.