Sustainable Living through Ambient Intelligence: A Bi-directional Publish-Subscribe System Using a Cross-layer Design

Abstract

Event Abstract Back to Event Sustainable Living through Ambient Intelligence: A Bi-directional Publish-Subscribe System Using a Cross-layer Design Yang Liu1, Boon-Chong Seet1* and Adnan Al-Anbuky1 1 Auckland University of Technology, Department of Electrical and Electronic Engineering, New Zealand The current research on ambient intelligence (AmI) has focused largely on enhancing the healthcare of people, such as the elderly and disabled, by supporting their daily activities through AmI. Recently, AmI has also been applied in the persuasion of people into behavioural changes, such as to behave in a more sustainable and environmentally responsible manner through delivering persuasive content tailored to users at the right time and at the right place [1]. In this paper, we propose to move beyond persuasion to actually support people towards an ecologically sustainable way of life through actionable intelligence, which we define as the timely information provided by the ambient environment that allows the user to initiate a course of action for more sustainable outcomes.For instance, sustainable living entails a conscious effort of the people to use water and energy resources more efficiently. Future textile may be smart tagged, e.g. using RFID, with details on the type of textile, who it belongs to, and embedded with sensors such as E-nose to detect its soiled level. To enable water- and energy-efficient laundering in an ambient home environment, the actionable intelligence to the user can be the auto-identification of all soiled items in the home to maximize the wash load, and also their type and soiled level to allow choosing the most efficient settings for the wash machine. Reducing production of waste and using products that are more environmentally sustainable is another approach to reducing our ecological footprints. To support the notion of ‘sustainable shopping’ where wastage of food and use items is kept minimal and ‘eco-labeled’ products are given greater consideration, the actionable intelligence to the user can be the current usage level of use items and consumption level of food as detected by, e.g. embedded weight sensors in item holders, which in turn generates a list of recommended products that are eco-friendly while matching the cost budget of the user.In this paper, the publish-subscribe communication model (Fig. 1) is adopted for efficient, timely delivery of actionable intelligence from the ambient environment to users. In simplest form, publishers are just sensors that publish sensed data to broker nodes, which in turn deliver to users who have subscribed to receive them. Different users can subscribe to different types of data from different sensors. The publish-subscribe model is chosen as it does not require all entities of the system (publishers, subscribers, brokers) to be connected with each other, all at the same time. Thus, it can be better at handling common issues in sensor networks such as intermittent connectivity and sensor failures. This paper further proposes a novel bi-directional publish-subscribe system, where publishers can also be subscribers, and vice-versa, and thus information flow can occur in both directions. A context-aware user device (or appliance) that subscribes to certain sensed data to generate context at higher level of abstraction, can also be a publisher of this context, which can be subscribed by low level sensors for optimizing their performance (Fig. 2). Our future work includes a performance analysis of the proposed bi-directional publish-subscribe system and its support for cross-layer sensor network optimization using higher-layer context information. 312_Figure_2 312_Figure_1