Multi-Keyword Based Information Routing in Peer-to-Peer Networks.

IoT devices are dispersed over the edge of the Internet and, naturally, so does the IoT data. As we know, data are valuable resources and sharing of these IoT data has many benefits. Correspondingly, we develop the peer-to-peer information centric routing ( P2PICR) solution to enable the discovery of IoT data sources and, subsequently, facilitate sharing of the IoT data on the edge. In P2PICR, keywords are used to index the routing tables to support search query routing toward annotated information sources. Though many DHT (distributed hash table) based solutions have been developed for managing p2p data, they are not suitable for IoT data sources. Thus, we consider the unstructured p2p routing as the basis for P2PICR. However, due to the large keyword space for specifying the IoT data sources, the p2p routing table has the potential of high space overhead. Thus, we develop a routing table space reduction scheme to significantly reduce the space requirement for the keyword-based routing tables. The routing table size reduction scheme is based on keyword summarization, which extends the IP address summarization technique adopted in IP routing tables, but keyword summarization is far more challenging than IP summarization. Compared to the conventional unstructured p2p routing in which the keyword-indexed routing table entries are cached and being elicited when the routing table is full, our summarization scheme instead compress the entries to retain its information for guiding the routing process. We conducted experimental studies and the results show that P2PICR achieves 20-30 times of routing table size reduction when compared to the case of without summarization and achieves 2-3 times communication cost reduction when compared with cache based solutions under the same table size.