An Improved DG Model for Prediction

The key of prefetching techniques is to establish an effective prediction model, which is mainly classified into two main categories according to the data structure namely PPM model and DG model. However, both models have their own shortcoming. To overcome their shortcomings, we will propose an improved algorithm based on exponential descendent double dependency graph algorithm. And we will also optimize the model in breadth which obeys Zipf’s distribution and in depth which obeys Gaussian distribution. Experimental results show that the model achieves certain effectiveness in improving prediction accuracy and reducing space complexity.