Ambient Intelligence: A review by Alex Bird
Part 1
Invisible technology, embedded in our natural surroundings, present whenever we need it, enabled by simple and effortless interactions attuned to all our senses, adaptive to users, utilising data whilst preserving privacy. This is Ambient Intelligence.
The devices used in this environment aim to be low weight, low power and low cost with an unlimited supply. The additional technological achievement that is required for this kind of environment aims to be achieved in the next 20 years. This book was published in 2005 so Weber predicts for this advancement to be reached by 2025.
The reasoning behind the development of Ambient Intelligence is to provide people with better control of their surroundings and personal support from these ‘invisible’ systems to create a better quality of life.
Part 1 of the book introduces the idea of Smart objects. These objects, just like humans, can explore their environment, communicate with other Smart objects and interact with humans. However, integration of these objects have so far proved difficult and hard to predict.
A term introduced is Ubiquitous computing. This idea means having computers running in the background for people, freeing up tedious routine tasks. In the future, a ‘new economy’ sees wireless communications attached to everyday objects. These products will be ‘self-conscious;’ washing machines that automatically detect the washing instructions on your clothes, a car that lists its maintenance history itself, a printer that orders its own printer paper.
However this ‘new economy’ comes with some issues and it will be these issues that hinder the acceptance and therefore complete emergence of Ambient Intelligence into our world.
Is it right that we might not notice technology running in the background monitoring us?
Will people be happy with a fully computerized society?
Will this promote the disappearance of nature?
Will people accept this change in how we live?
Will there be too many disturbances from technology?
Will some interventions encourage an even more consumerist society?
Will there be more opportunities for manipulation through advertising?
Who will control the systems and the content involved?
Will technology fail and become more of a hindrance than a help?
Could there be some long term health issues being amongst electromagnetic radiation?
Will these new devices be biodegradable or recyclable?
Weber explains 4 borders of privacy; natural, our very walls and clothing, social, privacy between groups of people, temporal, privacy about phases in our lives and ephemeral, privacy about fleeting moments. Considerations with what is kept private and public will alter our future social and legal norms and a very cautious approach is needed, to avoid killing off the idea of ambient intelligence altogether before it’s even begun.
An experiment called Homelab was studied to test some Ambient Intelligent systems. A test home was set up and a light system was used to enhance the experience of films and music in the home. The user response was that the system felt too alien to them and that the system was not natural or embedded well enough into the environment. However, in 2020 there are now systems in place like this that are quite successful.
Part 2
Part 2 discusses Networked Infomechanial Systems. These are systems made up of elements that need to be connected to each other and untethered. Two parts in these systems include Actuators, a type of machine, and Nodes, which are intersection points. These systems are used for sensing and sampling natural, indoor and urban environments. For example, these systems can be deployed in natural environments for monitoring natural disasters. They can also use image sensors to create images of a forest ecosystem. These systems can run into uncertainties and unpredictability, for example, changes in vegetation growth, having effect on acoustics in an environment, or new or foreign objects that block sensors.
Tiny OS is a base platform used for sensor networks. Tiny OS uses ‘NesC’ language to programme inter component communication, like ‘commands’ and ‘events’, and individual components like ‘tasks’. For example, one command function will be ‘getData(),’ and an event function will be ‘dataReady().’ Tiny OS was made to meet 4 requirements; limited resources, reactive concurrency, flexibility and low power or sleep mode. Therefore, Tiny OS is ideal for situations like habitat monitoring.
A trial was tested where the system was set up on an island and connected to the internet via satellite to a base station. Sensors would stay in sleep mode to preserve energy and then wake up to pick up data every 68 seconds. 1.2 million sensor readings were recorded over 4 months. The system was charged with solar panels. There were no failures in the system and all the data needed was collected, so it was considered a success.
Next Weber introduces ‘Ad hoc Wireless Sensor and Actuator Networks,’ (AWSAN). These networks sit in the background until they are needed. They are a connection of sensors, controllers (which use algorithms) and actuators, that together monitor and control the environment they are embedded into. A ‘demand response’ system uses this network. Eg. a system connected to a home thermostat will monitor energy use, temperature, price and weather and then display a ‘traffic light’ system, signalling to the user when prices are high and low. The user can choose to turn the thermostat on or off depending on their cost and usage preference. These networks rely on being simple, flexible and interoperable enough to be a success. Systems like these will also require the nodes in the system to automatically localise and synchronize with each other.
The most prominent challenge attached to these networks is security. The many devices and communication channels mean a security attack could happen anywhere. Unlike modern day computers that have, for example, firewalls, separating inside from outside, network systems mean the user and the attacker are both on the inside. VPNs are not suitable as they would be too expensive to set up on each individual component. The system is only as secure as its weakest link. However, attacks on particular nodes may not be overly harmful as obtaining one piece of information from that node might not be so helpful to the attacker, as the whole system is not disclosed. Some examples of system attacks include, ‘denial of service,’ which involves the attacker flooding the system with requests draining the energy and killing the nodes, ‘sleep deprivation torture attacks,’ which trick the nodes to go into sleep mode and ‘eavesdropping,’ where the attacker can extract information from conversations between nodes.
Some ambient intelligent scenarios are as follows:
Peter’s electric, gas and water usage is registered in an ambient intelligent system which connects to appliances in his house. The smart heater detects a fault and notifies the system. He is billed automatically based on precisely the energy he uses, through the online banking software.
Ginger’s embedded home system measures the air temperature and regulates it automatically using controls for the air con and windows. Insulation detectors get information about solar radiation and control the blinds accordingly. She also has ‘electronic gardeners,’ which notifies her when to water the plants.
Grandpa Jim no longer needs to live in a care home because he has a system that cares for him and monitors his health at home. When he leaves the house the system locks the door behind him. His ‘smart shirt,’ interacts with the ‘smart road signs,’ to communicate with Jim and give him directions around town.
Part 3
Weber introduces Moore’s Law. Gordon Moore, the co-founder of intel, wrote a law about the continuous enhancement and improvement of technology by momentum. The law predicts that, for example, costs will continue to fall, storage will get bigger, wireless speeds will get faster, etc. This is true as of today, however, technology particularly for sensors, actuators, ultra low-power radio and smart materials, will need to advance more, for Ambient Intelligence systems to be properly deployed. The driving factor for these technological developments depends on fashion trends. However, Weber warns us that technology should not bombard the consumer but help them, keeping people at the heart of the design.
Next, Weber discusses processing. Importantly, processing takes up less energy when the code is simpler and there is less of it. Moore’s law states that naturally, code will shrink over time as long as it continues to be developed. Asynchronous logic, logic that is non-sequential or independent from other logic, is used to optimise storage and power. Three types of devices use this type of logic. ‘Autonomous micro devices,’ are the smallest devices, cheap and lightweight and used without supervision. They are less than 1cm in size and are placed in environments to monitor them sporadically, using sleep mode otherwise. ‘Portable mini devices,’ are workhorses, small and portable pocket sized controls, that use sensors, microphones and speakers. ‘Static maxi devices,’ are the largest in terms of processing and storage, they have no power constraints, and come in the form of a ‘home server,’ or large plasma displays. These have access to personal databases and allow user control over the ambient environment.
Algorithm development is important to bridge the gap between existing algorithms and those needed for ambient intelligent systems; in such fields as speech recognition, object recognition, intention based searching and reinforcement learning, among a multitude of others.
Although the full deployment of an ambient intelligent society was not reached by 2010 as predicted, and there was the creation of the ‘Apple Newton,’ the ‘Berkley Infopad,’ and ‘palmtops,’ now, in 2020, there has been a step up in technology that hint to an Ambient Intelligent future. For example, Siri, Alexa or Echo dot, which respond to voice and aid users day to day life; contactless payment devices, online banking and smart apps for the home that control, for example, heating; and the Sonos home sound system which involves embedding speakers around the house etc. However, most devices we use today are not fully embedded and invisible.
Weber discusses the importance of evolving power technology to enhance ambient intelligence systems. Network nodes will be implemented based on appropriate cost, size and power and for this we need to avoid expensive technology, have a minimal number of components, a high level of integration and simple and cheap packaging and assembly. Components must also be self-sufficient for the lifetime of the product and nodes need to be agile to start-up quickly from being in sleep mode.
‘Miniaturization’ technology reduces component size to embed components into our environment, provoking the potential extinction of our current screens, keyboards and mice. One idea embeds microchips into concrete structures so they can register new cracks or problems caused after earthquakes; or a ‘smart carpet’ that tracks people’s footsteps and registers intruders or emergency falls. There are technologies to reduce from ‘wafer thin,’ to thinner than a few micometers. ‘Smart dust,’ or an E-grain is a microchip smaller than a penny. It uses thin and flexible silicone created through size reduction methods such as thin film technology and lamination.
For wearable electronics there is ‘flip chip’ technology that uses methods of folding to make a reduced electronic package. Textile integration includes methods such as soldering, adhesive, joining, crimping, sewing and riveting. Some elements will be non-machine washable so there is integration technology in conductive velcro, push buttons and zips.
‘Intelligence’ refers to the social nature of the user interface and the extent at which systems can adapt and learn through interaction. Ubiquitous computing needs to exhibit emotion and detect moods for it to be ‘ambient intelligence’. A new search area called ‘affective computing,’ combines psychology with computer science. The interaction between man and machine should match that of human to human interaction.