Siva - The IPFS Search Engine

Recently, InterPlanetary File System (IPFS) has attracted considerable attention as a method to overcome the weaknesses of the current Web such as the single point of failure and an arbitrary control by the government. Each file stored in IPFS is associated with a unique cryptographic hash (CID) as a logical address to enable users to acquire the content of the file without being aware of the network address. In this paper, we design a decentralized search engine for IPFS which quickly returns a list of CIDs associated with a given collection of keywords. A key idea of the proposed method is to store the association between keywords and CIDs in the Distributed Hash Table (DHT) used in IPFS, and to utilize a result cache to accelerate the processing of partially duplicated queries. In addition, we use a variant of Bloom filter to quickly check the availability of the result cache. The performance of the proposed method is evaluated by simulation. The simulation results indicate that it takes 1.8s on average to respond to a query in a network of 1000 and 2000 nodes reduced by 83%, 81% respectively when the result cache is added to the search engine.