The future Internet of Things (IoT) demands a closer breach between fixed and mobile envi- ronments in the content plane; so, even if devices are loose coupled the information they produce should be available from end-to-end. In this context, the Publish/Subscribe (Pub/Sub) paradigm  has emerged as an attractive communication model for exchanging content be- cause of its asynchronous, time and process decoupled style. On the current Internet, the Pub/Sub systems have been successfully extended to content dissemination services such as: PubSubHubbub  as well as they have being considered as a promising content-centric communication model for Future Internet Architectures . Despite this, there are still challenges concerning how to deal with the heterogeneity across entities that produce and consume content. Some of these entities can be mobile devices with low performance, non-stable connectivity and tight energy constrains; thus, their communication capabilities largely differ from fixed devices (e.g. a server in a datacenter).
Publish/Subscribe systems, which are also knows as event-based systems, are basically com- posed of three main components  publishers which are the content producers, subscribers that express their willingness to consume specific content; and finally brokers that put in contact publishers and subscribers by storing and forwarding this information. Depending on the scenario, wireless devices can perform as publishers, subscribers or brokers; so their ca- pability of consuming, publishing and matching content (in the form of events) depends on the expressiveness  of the subscription language of the Pub/Sub network.
In future IoT scenarios, it could be the case that even if a mobile device is capable of per- forming as a broker (e.g. a mobile phone, laptop, or medical wireless handhelds), it could still lack of network and processing capabilities in comparison with fixed devices. In addition, as these devices could appear and disappear at any moment they should perform as modular and pluggable components. In this paper we propose solutions that target these scenarios in the form of subscription distribution, notification models and hot-topic classification.
For full text: click here
(Author: Augusto Morales, Tomas Robles, Ramon Alcarria, Edwin Cedeño