A Partition Matching Method for Optimal Attack Path Analysis

Optimal attack path planning has a significant impact on network security. According to an optimal attack path, an attacker can quickly and efficiently attack the target host. This article proposes a partition matching method (PM) to infer the optimal attack path. PM can avoid the problem of path loss and get the result quickly. PM has two steps: network partition and local path matching. The target of the network partition is to divide the network into certain parts and find out the key nodes. The local path matching aims to splice the local paths in partitioned networks into a full path. In the target network, every host has a weight matrix which includes host connect number, CVSS value and their respective priorities. A cost function is proposed to calculate evaluation values of optimal local paths and the evaluation value of full optimal attack path based on the weight matrix. Results of our experiment demonstrate the capabilities of PM which can generate an optimal attack path in one single run. The results obtained by PM show good performance and are compared with other methods.