RELOAD Extension for Data Discovery and Transfer in Data-Centric Publish Subscribe Environments

Abstract

Data-Centric Publish-Subscribe (DCPS) is an architectural and communi- cation paradigm where applications, also known as data-participants, ex- change data-content through a common data-space by means of publishers and subscribers. DCPS systems are focused on data-content instead of on data location, and allow for performing advanced content-based operations. Due to its focus on data-content, DCPS is especially suitable for deploying IoT systems. However, some problems must be solved to support big scale DCPS IoT deployments. First, it is necessary to define an addressing scheme flexible enough for covering multiple DCPS-based IoT use cases. Second, publishers and subscribers must discover each other (i.e., a rendezvous func- tion is needed), perhaps in the presence of firewalls and NATs. In this paper we define a novel extension to the IETF REsource LOcation And Discovery (RELOAD) protocol specification for providing content discovery and trans- fer in big scale IoT deployments. We have conducted a set of experiments over multiple simulated networks of 500 to 10000 nodes that demonstrate the viability, scalability, and robustness of our proposal.