Decision model of intrusion response based on markov game in fog computing environment

For the current problems of complex network state, difficulty in fast response to intrusion and poor adaptation of response decision in fog computing environment, in this paper, we propose an intrusion response decision method based on a combination of deep learning and reinforcement learning based on game theory (Minimax-DQN). First, a Markov game is used as the standard to construct the intrusion response decision model in the fog computing environment. Second, two different sets of random variables are defined considering the possible behavioral choices taken by the attacker and the fog computing intrusion detection systems (FC-IDS), and the continuous state space formed by the game between the attacker and the FC-IDS is processed using a deep Q-network. Finally, the Minimax algorithm is used to solve the optimal value function in a specific state, and the best intrusion response strategy is obtained according to the training output after the training is completed. Three sets of experiments are conducted to compare the results of the Minimax-DQN algorithm, the DQN algorithm and the random strategy. The experimental results data prove that the model and the proposed algorithm can greatly improve the probability of IDS winning in the game process with the attacker, and thus effectively solve the problem of intrusion response decision in fog environment.