Design and analysis of an EPC discovery service based randomized architecture

摘要

Concept of Internet of Things has been proposed by professor Ashton from MIT Auto-ID for the first time, EAN and UCC established a none-profit organization EPCglobal, a lot of standards established to make the relative application canonical, aims to raise visibility in supply chains by enabling interested parties to query item-level data. EPCglobal defines EPC-IS, EPC-ONS, DTD, DTS etc. in its architecture but the standard of EPC-DS hasn't been proposed. EPC-IS stores the data of commodities which are delivered in supply chain, different companies own their own specific EPC-IS server and diverse data formats. The trend of Application scale is massive, data magnanimity and distributed, a lot of upper system services depend on historical data, so a reliable, high performance and efficiency data discovery mechanism is very important. Centralized discovery model is not suit for handling large number of requests, so it's a bottle-neck of application possibly; Chain discovery model exists risks of chain-breaking problems and single node failure, these problems could influence the stabilization of system; Net discovery model maintains the stabilization of system by increasing the auxiliary paths, but the complexity of space is too large, so it possibly doesn't suit for massive application; Distributed discovery model can not only maintains stabilization of system but also has a high performance and efficiency, Chord is a distributed mechanism which offers less complexity of space then Net model and gains a very considerable efficiency in average nodes searching. This paper propose a distributed discovery mechanism base on randomized architecture which offers performance of complexity of space as same as Chord and less cost of average nodes searching than Chord, and this mechanism can handles the chain-breaking problems and single node failure to offer a stable, robust discovery service.