AMQP and Web messaging for open and scalable distributed sensing apps
Future Web applications will increasingly require real-time data from the physical world collected by a myriad of sensors and actuators. Currently, integration of such devices require customized solutions due to the lack of widely adopted protocols for devices.
In such conditions, a high-level data-centric approach using such things as the Advanced Message Queuing Protocol - in which information is delivered based on content rather than explicit addressing of individual nodes- can simplify the development of applications that integrate data from these devices.
Although consumer electronics with Internet access are becoming increasingly popular, a common ground that enables seamless integration of devices with applications is still lacking. Indeed, the myriad of existing protocols and standards for networked devices turns each networkinto isolated islands that hardly interact with each other.
Because the Web is widely used, open, and easy to integrate, the Web of Things vision proposes to reuse the wide-spread Web infrastructure and standards (HTTP/XML/RSS) to connect embedded devices.
In this paper, we extend the Web of Things paradigms to support more scalable and interoperable WSN applications by leveraging the event-driven nature of publish-subscribe systems.
A priori, HTTP seems not suited for building WSN applications because of its request/response nature. However, the recent success of Web push tools and techniques have enabled the development of event-driven applications directly over the Web. Our results show that RESTful messaging for embedded devices is a viable and scalable approach for building more open and programmable distributed sensing applications.
Because the Web architecture offers a high degree of interoperability and a low entry barrier, we propose to leverage the Web to build hybrid applications that combine the physical world with Web content.
Our work builds upon recent developments in Web push techniques and extends them for embedded devices with a RESTful messaging system. Our results illustrate that fully Web-based distributed sensing applications are not only feasible but actually desirable because Web standards offer an ideal compromise between performance and functionality.
To our knowledge, our work is the first to show the feasibility of a fully Web-based distributed sensing application that combines the recent advances in Web technologies to design an HTTP-based event-driven programming model for sensor networks.
To read this external content in full, download the complete paper from the author archives at ETH Zurich.