Personal e-Comfort Modelling and Management based on

Multi-Agent System and Internet of Things Network

Benjamin Gateau

and Jarogniew Rykowski

SSI, Public Research Center Henri Tudor, Luxembourg, G-D. of Luxembourg

Department of Information Technology, Poznan University of Economics, Poznan, Poland

Keywords: Internet of Things, Multi-Agent Systems, Comfort Management, Ubiquitous Computing.

Abstract: In this paper a new approach to complex comfort management is presented, aiming in automatic treatment

of different comfort parameters by means of Internet-of-Things devices and multi-agent system. The paper

presents a new model of e-comfort, based on common treatment of all the parameters as identified across

the Maslow’s hierarchy of needs. Next, a new architecture of e-comfort management is discussed, based on

two layers: low-level layer of IoT devices, representing at-the-place possibilities of the system, and upper

layer of software agents, formulating and negotiating the needs and expectations of human users.

1 INTRODUCTION

Comfort is often provided in homes or places of

work by automation systems connecting actuators

(shutters, lights, heating, etc.) and sensors and

linking them together through scenarios and/or rules.

Typical scenarios define rules triggering the

execution of one or several actuators according to an

event. The event can be based on time (when it is

6:30 AM, execute the “wake-up” scenario), can be a

result of a direct interaction with the user, e.g., while

pushing a button (physical on the wall or virtual on

an interface) or the value reached by a sensor (when

the inside temperature is more than 25°C, switch on

the air conditioner).

The above notion of comfort is quite

straightforward and not deep enough. Comfort is not

only a set or rules or predefined scenarios, comfort

is equilibrium between the needs of the user and

his/her environment. Users want to be comfortable

without taking a lot of time to perfectly adjust

settings of their environment. The home automation

system has to anticipate the needs of the users in

order to reach a comfortable environment. The idea

behind is to automatically provide comfort at certain

level as Weiser’s good servant rules (Weiser 1991)

based on ubiquitous computing. This kind of e-

comfort and fully manageable conditions at-a-place

would be done via smart IoT devices able to

coordinate themselves to bring intelligent

assistantship and switch-less or hands-free home.

The goal of the paper is to propose a new way of

complex comfort management, having a set of

devices at one end, and human needs and

expectations at the other end. The idea is to manage

the comfort automatically based on the fuzzy-

declared parameters and available at-the-place

devices, to minimize manual activities of a user. The

paper is the first step towards such automatization,

aiming in the presentation of the way of comfort

modelling and controlling.

The remainder of the text is organized as

follows. In Section 2 we discuss basic comfort

elements that should be taken into consideration.

Then, Section 3 describes how these elements are

used to model the comfort in a complex and uniform

way. Section 4 briefly describes system architecture,

and Section 5 provides some conclusions and

directions for the future work.

2 COMFORT ELEMENTS

Comfort is surely not a simple and costless goal to

achieve. On the contrary, we see the notion of

comfort as extremely complex and embroiled. Thus,

we devote this section to a description and

discussion on the components of the whole term, in

division to physical conditions of the comfort,

psychological aspects, economics aspects, and

Gateau B. and Rykowski J..

Personal e-Comfort Modelling and Management based on Multi-Agent System and Internet of Things Network.

DOI: 10.5220/0005229900830089

In Proceedings of the 5th International Conference on Pervasive and Embedded Computing and Communication Systems (PECCS-2015), pages 83-89

ISBN: 978-989-758-084-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

safety/security conditions, these are enumerated and

briefly discussed below.

Due to the well-known theory of self-motivation,

people tend to live in so called comfort zone – our

everyday home&work environment. Comfort zone

determines our way of live, our friends and any

other social contacts, our customs, beliefs,

imagination, goals etc. Within our comfort zone, we

feel safe and secured, and we tend to keep this

situation stable. Sometimes we are forced to go out

from the comfort zone, to the adaptation zone –

there, we do not like the situation and we tend to

change it, keeping closer to the missing comfort.

This is a positive trend, as we are active and make

several efforts on inducing the changes – to us, or to

the environment. However, sometimes the change is

so critical we land in the panic zone – there, we are

not able to react to the changes and go back to the

comfort zone, and we need some help and

assistance.

Anyway, the overall goal of keeping ourselves in

the comfort zone is achieved by:

 understanding of what “comfort” means to every

one of us, and

 addressing some external assistance to help us to

go out from the panic and adaptation zones.

As mentioned before, we consider the comfort as

a balance between the user and his environment. We

are constantly evolving in changing contexts. By

context we mean “any information that can be used

to characterize the situation of an entity”. Here, the

entity is the user and the information that

characterizes the situation is the one that can be

compared to the user’s needs to be comfortable.

Thus, it is crucial to identify our needs and

expectations related with comfort, in as many its

aspects and possible, and explain these to the

assistants – humans and recently intelligent

devices/places – in order to improve the unwanted

situation. The first step towards such identification is

to classify all possible sub-notions of comfort, to be

personalized in the second step towards optimal

adaptation to individuals, situations, places etc. To

this goal, in this section we discuss several aspects

of the comfort, starting from the physical conditions

(natural and close environment), and finishing on

some psychological and social aspects.

According to well-known Maslow’s theory on

the hierarchy of human needs, our behaviour (and

indirectly – the comfort) is determined by a level of

fulfilling the needs. The needs create a hierarchy, to

be fulfilled in certain order. A need provokes certain

actions to fulfil it. Maslow's hierarchy of needs is

often portrayed in the shape of a pyramid with the

largest, most fundamental levels of needs at the

bottom and the need for self-actualization at the top

(Fig. 1) (Maslow 2014). Starting from the bottom of

the pyramid, physical conditions of the comfort are

related with our natural environment and its

measurable parameters – temperature, air flow,

humidity, lightness, etc. These conditions are

directly perceptible by the senses: sight, hearing,

touch, including sense of temperature, smell, taste,

balance and more. Thus, a natural trend is to link the

comfort aspects to the senses, as depicted below.

Figure 1: Maslow’s hierarchy of needs.

2.1 Thermal Comfort

Thermal control is mainly related with the sensitivity

of our skin towards the detection of external

temperature. Humans may exist and be functional in

a wide range of the temperature – starting from

-100°C (in clothes, or for a very short period,

however) to +100°C (for a very short period, or well

protected by special fire suit). The thermal comfort

zone is very special for particular humans, however,

most of us see an air temperature around 22°C as

optimal. Thermal control is strongly related with air

flow and drafts, temperature gradient, clothes, etc.

This topic was addressed many times by the

researches working in different domains, perhaps the

work (PhD thesis and further books) of Ole Fanger

is the most completed in that area (Fanger 1970).

Maintaining thermal comfort for occupants of

buildings or other enclosures is one of the important

goals of HVAC (heating, ventilation, and air

conditioning) systems – c.f. the description for air-

conditioning comfort below.

2.2 Ventilation and Air-Conditioning

Comfort

Addressed to closed environments such as buildings,

ventilating (the V in HVAC acronym) is the process

of "changing" or replacing air in any space to

provide high indoor air quality (i.e. to control

temperature, replenish oxygen, or remove moisture,

odours, smoke, heat, dust, airborne bacteria, and

carbon dioxide). Ventilation is used to remove

PECCS2015-5thInternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

unpleasant smells and excessive moisture, introduce

outside air, to keep interior building air circulating,

and to prevent stagnation of the interior air.

Ventilation includes both the exchange of air to the

outside as well as circulation of air within the

building. It is one of the most important factors for

maintaining acceptable indoor air quality in

buildings. Methods for ventilating a building may be

divided into mechanical/forced and natural types

(Ventilation 2014).

Complementary to the ventilation, air

conditioning is the process of altering the properties

of air (primarily temperature and humidity) to more

favourable conditions, typically with the aim of

distributing the conditioned air to an occupied space

to improve comfort. In the most general sense, air

conditioning can refer to any form of technology,

heating, cooling, de-humidification, humidification,

cleaning, ventilation, or air movement that modifies

the condition of air (McDowall, 2006).

2.3 Humidity/Hygrometric Comfort

Humidity is the amount of water vapour in the air.

Humans are sensitive to humid air because the

human body uses evaporative cooling as the primary

mechanism to regulate temperature. Under humid

conditions, the rate at which perspiration evaporates

on the skin is lower than it would be under arid

conditions. Because humans perceive the rate of heat

transfer from the body rather than temperature itself,

we feel warmer when the relative humidity is high

than when it is low. Some people experience

difficulty breathing in high humidity environments.

Some cases may possibly be related to respiratory

conditions such as asthma, while others may be the

product of anxiety. Sufferers will often

hyperventilate in response, causing sensations of

numbness, faintness, and loss of concentration,

among others (Humidity 2014).

Above-mentioned air conditioning reduces

discomfort not only by reducing temperature, but

also by reducing humidity. In winter, heating cold

outdoor air can decrease relative humidity levels

indoor to below 30%, leading to discomfort such as

dry skin and excessive thirst.

2.4 Gases and Smells/Odours

Smell sense is responsible for the detection of

certain chemical compounds dissolved in the air –

not only gases, but also liquids and solids.

Historically, this sense is the most archaic one – first

living organisms on Earth were equipment with such

detectors for communication and self-security.

Human smell sense is not as efficient as the one for

some animals, anyway, we are able to detect several

compounds, some of them are neutral for our health,

some of them (un)pleasant, and some dangerous,

especially at higher concentration. Unfortunately,

humans are not able to detect some very dangerous

compounds such as carbon oxide and dioxide. Thus,

broadening smell sense to detect all the unwanted

and dangerous components in the air is a very

desired part of the comfort zone.

Most heating, ventilation and air conditioning

systems (HVAC) re-circulate a significant portion of

the indoor air to maintain comfort and reduce energy

costs associated with heating or cooling outside air.

When occupants and building operators sense air

coming out of an air supply duct, it’s virtually

impossible to judge how much of this air is simply

re-circulated air and how much is outside air.

Current technology allows easy and relatively

inexpensive measurement of carbon dioxide (CO

)

as an indicator to help ensure ventilation systems

(for high density occupancy zones) are delivering

the recommended minimum quantities of outside air

to the building’s occupants (Prill 2000). CO

is a

natural product of human respiration whose rate can

be predicted based on an occupant’s age and activity

level. Beginning as early as 1916 (Mechanical

Engineer’s Handbook by McGraw-Hill) and found

in the New York City Building Code of 1929, CO

of 800 to 1,000 ppm and 1,000 ppm respectively

were recommended. However, the key point is that

levels are good predictors or surrogates for

human emitted bioeffluents (i.e., odours) that are

considered undesirable for the overall human

comfort inside conditioned spaces. Thus CO

is a

surrogate for levels of other bioeffluents that cause

odours that are likely to be viewed as unacceptable

by others in the space, not because of their presence

as a direct health hazard (Petty 2014).

2.5 Visual and Light Comfort

Light, mainly detected by the sight sense, is crucial

for human feeling of the comfort. Humans used to

act in the day-night cycle, with sunlight marking the

period of the activity, and darkness indicating the

period of the rest. Both are needed for our health,

and both may vary and depend on very individual

preferences, including the mood, company, etc.

Global tendency is to achieve the level and quality

of the light as close to natural (sun, outdoor) one as

it is possible (thermal temperature and spectrum,

distributed sources and background light, lightness,

Personale-ComfortModellingandManagementbasedonMulti-AgentSystemandInternetofThingsNetwork

etc.). Historically, lights and lightness control is the

fundamental issue for any modern (not to say

“intelligent”) building, starting from simply

dimmers, via light scenes, to nowadays LED-based

“walls of light” with distributed sources of the light,

fully controlled with respect to spectrum, intensity,

direction of the light beam, etc. Thus, a lot of

research has been already devoted to this aspect,

resulting in plenty of systems, norms, regulations,

standards, etc. Anyway, optimum lighting is crucial

for good sense of the comfort.

2.6 Acoustical Comfort

Noise and sound background is an essential element

of our life. It is hard to imagine we cannot hear

anything at the moment, thus for most of people

such silence is not a comfortable situation. On the

contrary, too much noise is also not needed, as it is

usually very disturbing.

Source of noise are internal of the

home/building, inside the same room or from others

rooms or could come from outside of the

home/building. It is influenced by first of all the

insulation used for external wall and the windows

from one side and for the internal walls and door

from other side. The user can act on the environment

to reduce the noise as, for instance, closing doors

and windows.

For those who do not like too much noise – there

are basically three ways to avoid it. The first one is

passive: walk away of the noise emitter. Surely, this

simple approach is not efficient in many cases, such

as a small room or a crowded place. Second

approach is to do some simple things to reduce the

noise, such as closing the windows mentioned

above. And third, we may apply some advanced

technical solutions, for example active headphones.

Such a device generates sound of the same intensity

but in opposite phase, thus somehow enduring

source of noise. It is questionable if such generation

is healthy or not, but in some cases it cannot be

avoided (such as ear protectors for ground handling

staff at an airport, construction workers, etc.).

By contrast, user could like to have a soft

background noise as classical or relaxing music for a

studious and focused environment or loud music. It

depends on the user’s needs and preferences but also

on the type of activity he is doing. For those who do

not like silence – this condition is correlated with

background “noise” such as music/TV programme.

2.7 Other Comfort Aspects

The upper is level of the Maslow’s pyramid of needs

the more the needs are abstracted and related with

psychology and sociology rather that physical

conditions. Thus, it is quite hard to maintain these

parameters by means of IoT devices. Anyway, some

of them should be taken into consideration for

complex comfort management, to enumerate at least:

 societal acceptance and the need for efficient

communication with other people regardless

place and time (mobile phones and VoIP

communication tools well serve to this goal),

 leisure, free-time organization (even if seen by

most of people as an evening in front of a TV),

 fear for bad activity, self-learning of new

environments (including RTFM problems for

new devices/technologies at home and work),

 stable economic situation,

 safety and security aspects, including personal

security, alerting and anti-theft systems, etc.

3 COMFORT MODELING

As drawn from the list of comfort elements given in

the previous section, there are two basic types of

comfort aspects: those that reflect the (desired) state,

and those that are a consequence of a fulfilment of a

need. For the first group, we cannot simply express

what we really want, but being in certain situation

we may usually assess if we like it or not. For the

second group, we are able in advance to define

several strict demands, and further we can measure

how far we are from the fulfilment of these

demands. Thus, we may say that the notion of

comfort is a summary of two aspects (Fig. 2):

 contentment, that could be define as a state of

being contented, or happiness in one’s situation,

 satisfaction, that could be seen as a fulfilment of

a need (a desire).

Figure 2: Modelling e-comfort as a summary of

contentment and satisfaction.

The contentment in one’s situation is however

PECCS2015-5thInternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

linked to the satisfaction of needs and is a function

of factors. For instance, the International Standard

ISO 7730:2005 (ISO 2014) says about the thermal

comfort: “A human being's thermal sensation is

mainly related to the thermal balance of his or her

body as a whole. This balance is influenced by

physical activity and clothing, as well as the

environmental parameters: air temperature, mean

radiant temperature, air velocity and air humidity.”

So the fact that a human feels well in a certain

situation is not only due to a fulfilment of a need

relating to the air temperature, and some other

parameters should be taken into account. These

parameters composed the context in which the user

is situated, namely air velocity, air humidity etc. If

we generalize this idea to all comfort’s components,

we can see that e-comfort cannot be only a set of

values setting the needs. If we take another example

related to luminosity, user’s needs cannot be

restricted to light source and light intensity only, in

all situations and places. Indeed, if the user prefers

natural light as a light source, and the night comes,

the user clearly cannot be fully satisfied. His needs

are dependent on the time, the location and the

activity (to mention a few basic factors). Similarly,

if he is at work, at home or at sport place, the needs

in term of light don’t will be same. As he tends to

read, cook, eat or watch TV, his needs evaluate

towards these activities. At a sport place, some of

the factors, such as luminosity, are not so important,

while a temperature takes a role of a leading factor.

In general, human preferences towards comfort

strongly depend on the environment, or globally - a

context, which constraints user’s needs according to

(among others) time, location and activity.

Dealing with the context is not a trivial task.

Returning to the previous example, the user should

be able to define needs composed by an ordered

choice of light source and luminosity value (if a light

source is the first choice, no matter the luminosity

value, he/she will prefer natural light, or, if the first

choice is related with an amount of luminosity, no

matter from where the light comes, as long as there

is enough light, the situation is seen as comfortable),

linked to a contextual triplet {time * location *

activity}. But even if we can strictly specify

preferences for the same time, the same activity and

the same location, how can we define a moment of

optimum switch from natural light to the electrical

one because of bad weather suddenly bringing dark

clouds?

As we saw, the configuration of needs for the

user could be very difficult. Our goal is to infer

situations contenting the user from context. As

define in (Naudet 2011) “a situation is a known and

pre-determined context which is composed of

elements having a location and an interval of time

and being measured (having value) coming from a

sensor. A context concerns an entity (a user), having

a status and an activity”. From that we deduce our

representation of e-comfort (cf. Fig. 3) being an

evaluation/measurement/assessment of a situation

regarding needs (or needs regarding a situation).

Situation and needs are composed of physical

comfort components which complete the

specification of a context based on (at least) the

location, the activity and the identity of the user

besides time period and/or current time.

Figure 3: e-Comfort representation for a user.

4 TWO-LEVEL ARCHITECTURE

The general architecture of the system aims at

controlling e-comfort at two layers (Fig. 4). The

upper layer is based on a multi-agent system

organized with the Moise

Inst

model (Boissier 2007).

This level is responsible for comfort management,

Figure 4: Layered system architecture.

Personale-ComfortModellingandManagementbasedonMulti-AgentSystemandInternetofThingsNetwork

including the storage of user-defined requirements to

personalize the meaning of e-comfort (their needs)

and their processing regarding the situation.

Comfort definition is represented by software

agents in the form of fuzzy-assessed goals

constrained by a set of value used as arguments and

stored in the needs database. Goals are specified

through the Functional Specification of Moise

Inst

. As

the goals are restricted by the capabilities of local

IoT devices and thus usually to be reached partially

only, the system places the users in “the most

comfortable situation as possible”, however, not

necessary ideal from individual point of view. This

solution is reasonable and similar to the one widely

used for non-automated systems (“I’ll do my best,

but I cannot promise a lot”).

The meaning of comfort is mapped by the agents

to the requests for activation of IoT devices. The

requests are addressed to the lower layer. The IoT

devices are never addressed individually. Instead,

the requests are mapped to an activation of a set (a

conglomerate) of these IoT devices that are the most

suitable to fulfil this request. For that, the Structural

Specification of Moise

Inst

defined roles played by

agent representing IoT devices able to influence a

specific category of needs (temperature for

instance). The set of activated devices depends not

only on the request, it is also adjusted according to

the number and type of devices accessible at given

location, and independent context of invocation. The

above process is controlled by SITE (Semantic

Internet-of-Things Environment) (Rykowski 2011)

framework, originally designed to control dynamic

and mobile IoT systems. SITE made it possible to

propose an “intelligent” network capable of

addressing functions (services) provided by the

devices (including personal devices such as

computers/ smartphones/tablets etc.) rather than the

devices themselves, based on the semantic

descriptions of devices’ capabilities. It abstracts the

way the devices are reached through gateways and

other hardware able to communicate with the

devices according to different protocols like Zigbee,

EnOcean, Z-wave or even “traditional” WiFi, IR,

Bluetooth, etc.

5 CONCLUSIONS

In this paper a new approach to complex comfort

management is presented, aiming in automatic

treatment of different comfort parameters by means

of Internet-of-Things devices and multi-agent

system. First, we discussed comfort parameters,

based on Maslow’s pyramid of needs. We

concentrated on these parameters that may be

controlled and managed by means of IoT devices:

temperature, ventilation, humidity, noise and

acoustic background, communication, safety and

security, etc. Next, we propose a uniform way to

model the comfort pointing out mutual dependency

of several parameters (such as temperature and

humidity), and overall context (such as place and

time, but also a company of other people with

different needs and expectations). To solve possible

conflicts, we proposed two level architecture aimed

in indirect control of IoT devices (SITE

environment, lower layer) based on an ontology of

device functions, and negotiations by means of

software agents, being representatives of their

human owners (upper layer). Currently, we work on

the implementation based on SITE and Moise

Inst

MAS environments, within the scope of GOLIATH

(Goal Oriented Layered system for Interoperable

Activities of Things) project.

ACKNOWLEDGEMENTS

This work is supported by the GOLIATH project

jointly funded by the Poland NCBR and

Luxembourg FNR Lead Agency agreement, under

NCBR grant number POLLUX-II/1/2014 and

Luxembourg National Research Fund grant number

INTER/POLLUX/13/6335765.

REFERENCES

Boissier O. and Gateau B.: "Normative Multi-Agent

organisations: Modeling, Support and Control", in

proc. of Dagstuhl Seminar 07122, 2007.

Ergonomics of the thermal environment, International

Standard ISO 7730:2005, from http://www.iso.org/iso/

catalogue_detail.htm?csnumber=39155, 2014.

Humidity, Wikipedia note, from http://en.wikipedia.org/

wiki/ Humidity#Human_comfort, 2014.

Naudet Y., Reconciling context, observations and sensors

in ontologies for pervasive computing. In Sixth

International Workshop on Semantic Media

Adaptation and Personalization (SMAP), pp. 3–8,

2011.

Maslow's pyramid of needs, Wikipedia note, from

http://en.wikipedia.org/wiki/Maslow%27s_hierarchy_

of_needs, 2014.

McDowall, R., Fundamentals of HVAC Systems, Elsevier,

San Diego, 2006.

Petty S., Summary of Ashrae’s position on carbon dioxide

(CO

) levels in spaces, from www.eesinc.cc/

PECCS2015-5thInternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

downloads/CO2positionpaper.pdf, 2014.

Fanger, P.O., Thermal Comfort Analysis and Applications

in Environmental Engineering. McGraw-Hill, 1970.

Prill R., Why Measure Carbon Dioxide Inside Buildings?,

from http://www.energy.wsu.edu/Documents/

CO2inbuildings.pdf, 2000.

Rykowski J., Hanicki P. and Stawniak M.: “OSL

Language to Represent and Interpret Conglomerates of

IoT Devices Accessed by SOA Services”, in SOA

Infrastructure Tools: Concepts and Methods, Eds. S.

Ambroszkiewicz, et al., Poznan, 2011.

Ventilation (architecture, Wikipedia note, from http://en.

wikipedia.org/wiki/Ventilation_(architecture), 2014.

Weiser M., "The Computer for the 21st Century" -

Scientific American Special Issue on

Communications, Computers, and Networks, 1991.

Personale-ComfortModellingandManagementbasedonMulti-AgentSystemandInternetofThingsNetwork