A First Approach to Detect Suspicious Peers in the KAD P2P Network

Several large scale P2P networks are based on a distributed hash table. They all suffer from a critical issue allowing malicious nodes to be inserted in specific places on the DHT for undesirable purposes (monitoring, DDoS, pollution, etc.). However, no study so far considered the actual deployment of such attacks. We propose a first approach to detect suspicious peers in the KAD P2P network. First, we describe and evaluate our crawler which can get an accurate view of the network. Then, we analyze the distances between the peers and the contents indexed in the DHT to detect suspicious peers. Our results show that hundreds of KAD entries are clearly under attack during our measurements.