A peer to peer social networking service exploiting triangle relationship among friends

These days popular social networking services (SNS) are provided as a centralized database service storing all the data at data centers. Due to the privacy concerns of such centralized systems, Distributed Hash Table (DHT) based P2P SNSes have been proposed, where data query can be done through DHT routing. However, DHT based systems have their own problems. First, DHTs are optimized for universal data dissemination among arbitrary users so that it cannot exploit specific characteristics of SNSes. Furthermore, the frequent data splitting and shifting for the node join/leave operations of DHTs becomes a critical overhead. To tackle the problems of both the centralized and the DHT based systems, in this paper, we exploit the intrinsic characteristics of the social networks. In our proposed system, the data is stored only at the originator and its one-hop neighbors, i.e., the consumers. For a friend without the data to retrieve the data, the friend simply sends query messages to its own one-hop neighbors, which are likely to be the one-hop neighbors of the data originator. This approach reduces the query latency, is highly scalable, and more importantly does not require any data splitting or shifting as the nodes join and leave. We show the performance of our system through extensive simulations over a social network topology collected from a real SNS.