Risk analysis of information-leakage through interest packets in NDN

International audienceInformation-leakage is one of the most importantsecurity issues in the current Internet. In Named-Data Networking(NDN), Interest names introduce novel vulnerabilities thatcan be exploited. By setting up a malware, Interest names can beused to encode critical information (steganography embedded) andto leak information out of the network by generating anomalousInterest traffic. This security threat based on Interest names doesnot exist in IP network, and it is essential to solve this issue tosecure the NDN architecture. This paper performs risk analysisof information-leakage in NDN. We first describe vulnerabilitieswith Interest names and, as countermeasures, we propose a namebasedfilter using search engine information, and another filterusing one-class Support Vector Machine (SVM). We collectedURLs from the data repository provided by Common Crawland we evaluate the performances of our per-packet filters. Weshow that our filters can choke drastically the throughput ofinformation-leakage, which makes it easier to detect anomalousInterest traffic. It is therefore possible to mitigate informationleakagein NDN network and it is a strong incentive for futuredeployment of this architecture at the Internet scale