A Specific Risk Evaluation System for Live Virtual Machine Migration Based on the Uncertain Theory

Table 1

DQN & uncertain integration for migration risk evaluation.

---

The core steps of DQN and uncertain integration for risk system

---

1. Initialize the DQN structure for risk at each level.

2. Configure the coefficient values such as learning rate and discount factor.

3. Choose the appropriate degree of discretion according to computing capacity.

4. Set the random selection method for action choice.

5. Create memory bank and start computing cluster.

6. Monitor metrics and store the transition.

7. While (step<=max_iter || accuracy is not satisfied).

8.  If step % 50 == 0.

9.   Assign the parameters to target network.

10.  Choose 10 samples from memory bank.

11.  Calculate the target value with actual feedback and discounted evaluative value.

12:  Obtain the loss and train the model with batch sampling data.

13.  Store the transition.

14.  step_counter+=1s.

15. Integrate three-level risk by searching maximum value from uncertain approach.

16. Evaluate migration decision by integrated ultimate risk.

---