Small logo of ETH main building ETH Zurich : Computer Science : Pervasive Computing : Distributed Systems : Education : Student Projects : Abstract

A Smart City Infrastructure (M)

Status: Abgeschlossen

Abstract—One reason why the Internet of Things (IoT) is not yet established beyond academic projects is the lack of integration with existing Internet applications and common standards. Building upon the IETF Constrained Application Protocol (CoAP), the thesis is devoted to design and evaluate an architecture for large-scale integration of the IoT into the Web. The central addressed issues are caching for scalable performance, transparent HTTP mappings, and access control, as they are required for a smart city infrastructure.

Background

A major building block of the IoT are constrained networks with lossy links, often called wireless sensor networks (WSNs). The IETF has recently developed protocols that are custom-tailored for this environment. Two of them are IPv6 over Low power Wireless Personal Area Networks (6LoWPAN), an adaptation of IPv6, and the Constrained Application Protocol (CoAP), a lightweight RESTful protocol that can be easily translated in HTTP for integration with the Web. The goal of the thesis is to examine and assess whether a large infrastructure of constrained nodes—like those that compose a smart city—can be built upon these standards. The topology and authority of these nodes will be influenced by the physical distribution of sensors in rooms, buildings, departments, etc. Hiding this complexity and the network constraints is a core incentive. Thus, the architecture will focus on (reverse) proxies that network between the CoAP origin servers, clients, and the Web components (i.e., HTTP clients and servers). Typically, the proxies will be placed at the edges of the LoWPANs. The proxy also becomes a protection for the inner LoWPAN: it limits visibility and access to the clients. Due to multiple network authorities and performance requirements, proxies may also be organized in hierarchies.

Objectives

The development of the proxy will be realized on the top of Californium, a CoAP framework written in Java. A lightweight and RESTful database, such as CouchDB, will implement the proxy caching. This solution should guarantee a short response time, manage various caching policies and store CoAP messages efficiently. The mapping between HTTP and CoAP is quite straightforward, hence the thesis will address the integration issues with the two protocols (e.g., not to break the end-to-end principle of this hybrid communication). For security and safety, the proxy will also be enhanced with a fine-grained access control mechanism (e.g., ACLs) based on several factors. These could be the category of the requested resource, the number of attempts in a time slot, or the state of an actuator, but also usual firewall rules.

For a reasonable evaluation, it is necessary to simulate the characteristics of a smart city in a testbed. Californium’s adverse layer will be used to generate traffic that reflects the properties of LoWPANs (i.e., throughput and timeliness depending on the node and hop count). These LoWPAN emulators will be used to stress the solution, which will be deployed in a real network.

The contributions of the thesis will be the elaborated architecture and a two-fold evaluation. For the quantitative analysis, various performance indicators will be measured by experiments: the impact (e.g., speedup) of the proxy on the network, the data freshness, and the cache hit rate depending on the policies. The results achieved will be compared with the literature, whether or not belonging to RESTful solutions.
The qualitative analysis, instead, requires the development of services showing the integration of physical objects into the Web through the proxy. Services such as pachube can demonstrate the mapping HTTP-CoAP: the requests from pachube's servers will be directed to the CoAP sensors. While, the mapping in the opposite direction (CoAP-HTTP) can be tested with an actuator (such as a LED display) subscribing to a feed (e.g., twitter or news).

Student/Bearbeitet von: Francesco Corazza
Contact/Ansprechpartner: Matthias Kovatsch

ETH ZurichDistributed Systems Group
Last updated December 10 2012 03:40:28 PM MET ko