A dynamic group-based scheduling method for low-latency anonymous communication systems

Low-latency anonymous communication systems have been widely used over public networks. However, they are susceptive to passive or active timing-based attacks. Traffic analysis or intentional flow marking based on timing characteristics of the network flows can both correlate the outgoing flows with the corresponding incoming flows, which makes the systems be penetrated and compromises the users´ privacy. Due to the low-latency constraint, techniques defending against such attacks often lead to a negative effect on the performance of the systems. In this paper, a dynamic group-based scheduling method is proposed to reduce the performance lost when deploying anonymous communication systems. For each incoming flow of the system, we take a fragment within a short time window as sample. First all flows are divided into clusters by the packets number of each flow within the window, then each cluster is further divided into groups by the weighted variances of inter-packet interval of the flows in the cluster. In this way, the flows with similar timing characteristics are gathered into the same group. Thus the system can deal with the groups respectively rather than handling all incoming flows all together. The groups can be dynamically adjusted according to the change of the packets number and inter-packet intervals of incoming flows. Simulation on offline data and experiments in the real network environment both show that our method can obtain better performance than simply manipulating all flows coming into the anonymous communication system as a whole.