Research And Implementation Of Plan Recognition Model Based On Action State Graph

Intrusion detection is a safe mechanism, which is just developed during these 20 years into a system of dynamic monitoring and preventing or defending the intrusion behavior. And it is a process of recognizing and responding intrusion behavior. From the character of this technology, we can find that one of the most important hotspots in the research field of artificial intelligence—intelligent plan recognition theory, which has the same characteristic, namely the process for plan recognition and response.This paper is based on the functional characteristics from above two aspects. By analyzing the specific detecting problems of intrusion, we firstly put forward the concept of the invasive event, the invasive process, the action state graph and the critical degree etc. And in an invasive process of an invasive event, this paper constructs a plan recognition model by adopting the action state graph. And it analyzes the presumable multi-target states that will be achieved by attacker. Moreover, according to calculating for every critical degree of every presumable multi-target states, it can realize the valid recognition to the final target state that attacker will achieve, which is affected by the invasive action of the attacker under the initial state at the current time-step. Besides, on this basis, the plan recognition model can send the information such as warning information to safety administrator automatically according to the valid judgment of the final target state. Therefore, it can realize the response process to the invasive action of the attacker.This paper adopts the analysis and design thought which face to object, based on the formal theoretical research of this plan recognition model. And it uses standard c# language to design the plan recognition system and the demonstrating platform of simulation of it initially for the intrusion detection problems specifically, and under the closed fabricated environment state, it does some necessary tests on this plan recognition system for the intrusion detection problems specifically. And it is easy to find that this plan recognition system put forward in this paper can response for the intrusion before intrusion has succeeded. It is a necessary supplement to the classical plan recognition theory, and it strengthens the protection capability of system itself, so it can make that the safety of system never be defeated.