QMigration.java

import org.cloudsimplus.qlearning.QLearning;
import org.cloudsimplus.qlearning.StateActionPair;

public class InterDatacenterMigration1 {
    // Existing code...

    private final QLearning<Integer> qLearning;
    private final double learningRate = 0.3;
    private final double discountFactor = 0.9;
    private final double explorationRate = 0.2;

    public static void main(final String[] args) {
        new InterDatacenterMigration1();
    }

    private InterDatacenterMigration1() {
        // Existing code...

        // Q-learning setup
        qLearning = new QLearning<>(learningRate, discountFactor, explorationRate);

        // Start simulation
        simulation.start();

        printResults();
        System.out.printf("%n%s finished!%n", getClass().getSimpleName());
    }

    private void createAndSubmitVms(final DatacenterBroker broker, final int[] vmPesArray) {
        final var vmList = Arrays.stream(vmPesArray).mapToObj(this::createVm).toList();
        broker.submitVmList(vmList);
        vmList.forEach(vm -> {
            vm.addOnMigrationStartListener(this::startMigration);
            vm.addOnHostAllocationListener(evt -> {
                int currentState = calculateState(evt.getHost().getPeList(), evt.getVm().getTotalAllocatedMips());
                int action = qLearning.chooseAction(currentState, getAvailableActions(evt.getHost()));
                performAction(action, evt.getVm(), evt.getHost());
            });
        });
    }

    private int calculateState(List<Pe> peList, long vmMips) {
        double totalAvailableMips = peList.stream().mapToDouble(Pe::getCapacity).sum();
        double utilization = vmMips / totalAvailableMips;
        return (int) (utilization * 100); // Simplified state representation (utilization percentage)
    }

    private int[] getAvailableActions(Host host) {
        // Example actions: placing VM on different PEs, or migrating VM to another host
        return IntStream.range(0, host.getPeList().size()).toArray();
    }

    private void performAction(int action, Vm vm, Host host) {
        // Example: action = index of PE where VM is placed
        vm.setHost(host).setPeNumber(action);
        
        // Example reward calculation (adjust as per your use case)
        double reward = calculateReward(host);
        
        // Update Q-table
        int currentState = calculateState(host.getPeList(), vm.getTotalAllocatedMips());
        qLearning.learn(currentState, action, reward);
    }

    private double calculateReward(Host host) {
        // Example: Reward based on CPU utilization improvement or migration action
        double currentUtilization = host.getUtilizationOfCpu();
        // Compute reward based on current and previous utilization
        return currentUtilization < HOST_UNDER_UTILIZATION_THRESHOLD_FOR_VM_MIGRATION ? 1.0 : -1.0;
    }

    // Existing methods...
}