An Integration Platform for Private Assisted Houses

Flavio Corradini

, Francesco De Angelis

, Barbara Re

, Emiliano Anceschi

, Massimo Callisto De

Donato

and Paolo Iddas

School of Computer Science, University of Camerino, Via del Bastione 1, Camerino (MC), 62100, Italy

Filippetti S.p.A and SmartSpace s.r.l, Via Marconi, 100/102, Falconara M.ma (AN), 60015, Italy

Keywords:

Ambient assisted living, Smart Environments and Housing, Home Care Monitoring Systems, Remote

Monitoring, Assistive Technology and Adaptive Systems.

Abstract:

A Private Assisted House aims to deﬁne a novel care model focusing on the changing needs of people to

promote active life expectancy. This raises the need of personalization in the design and development of

Smart Home, so starting from users requirements we stress the need of integration suitable to support such

changing requirements. In this paper we discuss Private Assisted House integration platform focusing on its

conceptual model and reference architecture. The platform is deﬁned around a set of smart objects managed by

a home gateway that communicate with a Cloud Center. This organization provide two kinds of processing: (i)

local to the house, and (ii) remote. The local processing involves events, triggers, commands and automations

managed directly for the gateway. The remote processing implies communication from the house to the Cloud

Center that can provide intelligence to the house using high-level applications that use data correlation to

perform speciﬁc tasks.

1 INTRODUCTION

Italy is one of the European countries holding high-

est ageing index rate. This implies an exponential

growth of the costs of care for the elderly in the fore-

seeable future. According to recent demographic pro-

jections in Italy, over-65s in 2020 will be 22.5% to

reach the threshold of 32.6% by 2065. Even more

signiﬁcant is the increase of the over-80s, which will

increase from the same period from 2.8% to 3.7% and

10% (source: www.demo.istat.it) and the perspective

of longevity and active elderly population is growing

quickly. At the same time people, at least in Italy, like

to spend their elderly time in her/his home, a place

where she/he can continue to carry out usual activi-

ties both in the case of independent living or life in

the family.

From this context rise the need to provide popula-

tion with an adequate home care and risk prevention.

To do that technology evolution is a driver and make

possible guarantee the same level of care according to

budget constraints given among the others by the eco-

nomical crisis. Technology is able to guarantee also a

certain level of personalization since elderly proﬁles

are different from person to person and at the same

time they change from time to time.

In this paper we present a software integration

platform suitable to support a personalized care model

in their private home with the beneﬁt of innova-

tive and content-based services made by user-friendly

technologies. This means that different levels of ser-

vice are enabled by a ﬂexible software integration

platform that is introduced with this work, and it is

adequate for people with different degrees of aging

and/or disability according to the DfA ”Design for

All” paradigm (Mace et al., 1990).

The platform is based on the concepts of Internet

of Things (IoT) (Gubbi et al., 2013) with the ability

to measure, infer and understand environmental in-

dicators in a private house while rely on the power

of cloud computing and big data technologies to pro-

vide a virtual infrastructure that integrates monitor-

ing devices and analytics tools. We based our model

on a platform that can integrate and coordinate home

automation, tele-care solutions and smart-objects re-

sponding in a personalized way to concrete needs of

automation, prevention, safety and communication.

This new class of Ambient Assisted Living (AAL)

technology will bring new capabilities such context

awareness, anticipatory behavior, user friendliness

and ﬂexibility (Nehmer et al., 2006). Moreover, we

extended our model with advanced monitoring func-

Corradini F., De Angelis F., Re B., Anceschi E., Callisto De Donato M. and Iddas P..

An Integration Platform for Private Assisted Houses.

DOI: 10.5220/0005436000450052

In Proceedings of the 1st International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AgeingWell-

2015), pages 45-52

ISBN: 978-989-758-102-1

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

tionality by applying a probabilistic framework based

on the Bayesian networks in order to further improve

behavioral analysis based on the data incoming from

the house.

The development of the platform was conducted

within the Pass (Private Assisted House Project)

project. It is part of an regional public ﬁnanced ini-

tiative for the development and implementation of

real life diffusion of Ambient Assisted Living systems

(Rossi et al., 2014).

The paper is organized as follows. Section 2 de-

scribes the development of the conceptual model. Af-

ter this Section 3 introduces the integration platform

reference architecture. Section 4 presents a real sce-

nario with reference to the solution in practice. Fi-

nally, Section 5 introduces relevant related work and

Section 6 draw some conclusion.

2 PRIVATE ASSISTED HOUSES

CONCEPTUAL MODEL

The conceptual model of the Private Assisted house

(PAss ) is characterized by a multilevel structure with

a strong decoupling in data processing, both in the

implementation of basic functionality directly avail-

able in the house, and in the implementation of the

advanced features through methodologies and tech-

nologies that are located at the cloud level. It is re-

ported in Figure 1.

Cloud

Center

User Interfaces

- Caregivers

- Medical associations

- Family doctor

Home

Gateway

Telemedicine

Platform

synchronous/

asynchronous

communication

Smart Objects

Figure 1: The main components of the Private Assisted

House.

The architectural model deﬁnes the following

functional components.

• Smart Objects

A smart object is an object able to describe its in-

teractions with the physical world (Kortuem et al.,

2010). It has both physical properties and infor-

mation related to them. In addition to this, it is en-

dowed with the ability to communicate with other

objects and with the environment to which it be-

longs with the aim to interact with other objects

and coordinate the execution of complex actions.

Under the PAss project several smart objects have

been developed including: doors and windows

with automation functionalities, a liquid-screen

window with programmable opacity, motion sen-

sors, light controls, temperature controls, wear-

able sensors for monitoring vital signs, etc.

• Home Gateway

The home gateway component implements the

business logic for the local management of the

house. It uses data coming from the smart objects

and from external entities such as, for example,

the Cloud Center (CC) or the telemedicine plat-

form. In the architectural model, this component

consists of an embedded system developed with

open source technologies.

• Cloud Center (CC)

The CC component implements a higher level

of processing that involve data coming from the

house. This allows storage, normalization and

analysis according to a logic that is not usually

available in the local environment of the house.

The component is built using technologies that en-

able the management and persistence of data like

in a typical big data scenario.

The conceptual model outlined above is the start-

ing point for the deﬁnition of an integration platform

within the PAss house. The house will be equipped

with a software platform that offers integration fea-

tures for domestic devices in a communication net-

work inside the house. This provides integration ca-

pability with the outside towards the CC that is able

to realize complex functionalities that usually are not

available within a classic home automation. The

platform objective is the implementation of welfare

scenarios that could evolve over time involving de-

vices and automation inside the house while exploit-

ing complex functionalities made available outside.

The presence of objects, home gateway and cloud

center respond to the needs to have a physical interac-

tion with user and two level of processing, one related

to the house with small, simple actuation and one

able to introduce in the house new conﬁguration over

time to respond to the changing needs of the users.

New conﬁgurations are built in the CC automatically

from algorithms speciﬁc for the disease/disability un-

der control or by a human physician that control the

house environment.

The high-level view of the integration platform

software revolves around ﬁve key concepts: (i) de-

tection, (ii) processing, (iii) actuation, (iv) interac-

tion, and (v) communication. The goals we want to

achieved is to provide an assistive environment with a

non-invasive approach. We offer a single system with

ICT4AgeingWell2015-InternationalConferenceonInformationandCommunicationTechnologiesforAgeingWelland

e-Health

Basic Home

Automation

Smart ObjectsSmart ObjectsSmart Objects

Neutral representation

API - Basic Functionalities

Real time processing

Storage

Big Data

Telemedicine

platform

Application / Frontend for

specific features

Application / Frontend for

specific features

Application / Frontend for

specific features

Detection

Processing

Actuation

Interaction

Communication

Driver interface protocols and gateway component

Communication

with the outside

Assistive

Technologies

Home Automation

System

Intelligence

User interface

Communication

Figure 2: The architecture of the PAss platform.

many different functions and the opportunity to in-

crease them through the applications deployed in the

Cloud Center located outside the house.

The Figure 2 describes the features that are imple-

mented in the platform through the use of components

based on open source frameworks or through develop-

ment from scratch during the project.

The lowest layers of the system are responsible

for detection and actuation capabilities. Here the sys-

tem integrate both legacy devices for basic automa-

tion (already commercially available), and new smart

objects (developed in the project) aimed to introduce

assistive technologies in the house. The former are

integrated using well-known protocols (i.e. KNX,

Bticino) while the latter are built to a higher level,

agnostic transport protocol, through the deﬁnition of

payload in textual format (using the JSON notation).

Both categories of objects must be represented in a

neutral manner with respect to the underlying proto-

col. In this sense, the object is generalized with re-

spect to the speciﬁc protocol in a neutral representa-

tion that is based on the functionalities of the object

(for example a switch will exhibit the functionality of

on/off, and so on).

In a higher level to provide the platform with a

neutral representation, we use some abstractions to

ensure independence from the lower layers and to en-

able the ability to interoperate. In this regard, the

platform will incorporate the Open Source product

Freedomotic (Freedomotic web site, 2014), a soft-

ware component that has a ﬂexible and scalable ar-

chitecture that can interact with best-known proto-

cols of building automation as well as custom solu-

tions. Freedomotic exploits modern enterprise inte-

gration models and architectures of distributed com-

puting, along with the APIs used for its extension. It

was chosen as a reference after a phase of scouting

for home automation technological platforms with the

requirement of extensibility, open source philosophy,

technological maturity and the ability to provide APIs

to other system components.

The Freedomotic component is placed side by side

with a telemedicine platform. Telemedicine allows

the query for medical information and the administra-

tion of devices that collect such information. The data

gathering is made using REST web services with pay-

load described using the JSON notation. We can reach

the following measuring instruments to assess vital

signs supported by the system: (i) BPM Blood Pres-

sure Monitor, (ii) PFM Peak Flow Meter, (iii) BGM

Blood Glucose Monitor, (iv) VSM Vital Signs Mon-

itor, (v) HWS Health Weight Scale, (vi) BCM Blood

Coagulation Monitor, (vii) ECG ECG Monitor, (viii)

PO Pulse Oximeter.

The communication layer is provided by the

MQTT (Mqtt web site, 2014) protocol. This provides

a ﬂexible infrastructure to publish and subscribe mes-

sages with the ability to discriminate them using their

topic. Each publisher sends content to speciﬁc topics,

while each subscriber retrieves them subscribing for

updates related to that topic. The platform will pro-

vided two kinds of processing: (i) local to the house,

and (ii) remote.

The local processing involves the use of a model

based to events, triggers, commands and automations.

The events are generated by the smart objects and de-

livered towards the integration platform where they

are related with the triggers in the system. The trig-

gers detect particular conditions from the events con-

tent. This activate commands to the actuations of the

house. In this sense, the automation expresses a cor-

respondence between one or more triggers and one

or more commands. At this level, simple domestic

automations are possible in the form of if-then-else

rules.

The remote processing implies communication

from the house to the outside to send information

about its state and retrieve commands to be executed.

It refers to what happens in the CC that provides in-

formation persistence to high-level applications that

can provide automation to the house. At this level

there are domestic automations whose implementa-

AnIntegrationPlatformforPrivateAssistedHouses

tion is deﬁned by a correlation of high-level data that

cannot be achieved in the home gateway.

The CC represents the intelligence of the sys-

tem enabled by big data storage and analysis fea-

tures. In particular, the open-source distributed

framework Hadoop (http://hadoop.apache.org/) based

on the well-known MapReduce framework (Dean and

Ghemawat, 2004) and the non-relational database

HBase (http://hbase.apache.org/) based on Google’s

BigTable (Chang et al., 2006) provide the ability to

retrieve data and correlation about the house and its

occupants. This provides the ideal knowledge base

for the realization of applications whose functionali-

ties are based on inferences and correlations. At this

level, the information retrieved from the house and

from the telemedicine platform are related to achieve

a speciﬁc assistive goal (i.e. check the normal every-

day life of an elderly at home).

The solution chosen in PAss is the use of applica-

tions based on Bayesian networks (Pearl, 1988) that

allow the platform to perform correlations between

incoming data from the house and other external com-

ponents (for example the telemedicine platform) in

order depict facts from which produce actuations to

be executed in the home to support such scenarios.

A Bayesian network is used to model a domain con-

taining uncertainty in some manner. In our case this

uncertainty can be due to imperfect understanding of

the domain or incomplete knowledge of the state of

the domain at the time where a given task is to be

performed. Indeed, the measurement process of the

physical quantities that deﬁnes the state of the system

is made by a distributed sensor network that act in a

asynchronous fashion: in every moment of time some

of this observation of the physical world will be char-

acterized by uncertainty.

Bayesian networks have been used since they are

a representation of a probabilistic model, which is

for us the reproduction of a probability distribution

over a set of variables to support the implementation

of advanced features for the house. This approach,

which combines statistical methods and artiﬁcial in-

telligence, it is a useful tool in many ways, including

the ability to simulate and replicate complex situa-

tions. Moreover, Bayesian networks are able to high-

light the structure of a phenomenon by means of an

intuitive graphical representation allowing the non-

expert in the ﬁeld to understand relationships. Using

this approach it is possible to learn information from

the data and, at the same time, to introduce in the anal-

ysis an expert judgment (medical, etc.).

Based on this approach, in the PAss house we are

able to correlate information about the state of the en-

vironment with information about the vital signs sup-

ported by the telemedicine system in order to state

what are the condition of normality when living in the

PAss house besides conditions that could lead to dan-

gerous state as explained in Section 4.

3 PRIVATE ASSISTED HOUSE

INTEGRATION PLATFORM

3.1 Reference Architecture

The architecture of the integration platform is made

by several components represented in Figure 3.

Smart Object Smart Object

Home gateway

Actuator Actuator

MQTT

Broker

Freedomotic message bus

Smart Object

Plugin

Actuation

plugin

Cloud

Center

Figure 3: Communication ﬂows of MQTT messages.

The system consists of ﬁve basic parts:

1. A network of sensors whose task is to collect data

and send them to the home gateway.

2. The home gateway based on Freedomotic, that

manage the house and the ﬂow of incoming mes-

sages from the sensor network and the CC.

3. A broker for messages that redirect the received

messages to the subsystems that require them.

4. The PAss database that is responsible to store data

from the network of sensors and the events de-

tected by Freedomotic. It maintains the local his-

tory of messages coming from the CC.

5. A CC that store and analyzes the collected data

from the house by using the Hadoop framework

and Bayesian networks. It generates the actua-

tions for the home gateway in the house.

The data streams managed by the system are es-

sentially two:

• In black arrows, we highlight the data collected

by the sensor network that are sent through the

broker messaging system to Freedomotic. Data

is analyzed and redirected in three directions: (i)

to the PAss database in the home gateway; (ii) to

the user interface that update the graphical objects

with the values collected by them; (iii) to the re-

mote CC.

ICT4AgeingWell2015-InternationalConferenceonInformationandCommunicationTechnologiesforAgeingWelland

e-Health

• In white arrows, we highlight the messages com-

ing from the CC implementation which are for-

warded to the home gateway system. The local

system update the graphic interface of the actua-

tors for which the CC sent information. The sys-

tem forwards the commands to the real actuators

inside the environment.

3.1.1 PAss Data Model

Within the platform for the PAss house there are two

main data sources. The telemedicine platform, which

is an autonomous system with its own data model and

the data model speciﬁcally developed for the smart

objects. This model is divided into three parts corre-

sponding to the three types of messages that can be

managed by the platform:

• Measurement Payload. The data sent from the

smart object follow a data model that is used for

the local processing in the home gateway, for the

update of the user interface, and for the CC. The

standard data that are sent from each smart ob-

ject contain: the date and time of the measure, the

type of smart object that carried out the measure,

the address associated with the object, its unique

identiﬁer within the network of sensors, a Freedo-

motic name and address. These data are generated

at each measure. The speciﬁc data collected from

the single sensor is sent as a tuple with the name

of the collected measure, its value and its mea-

surement unit.

• Actuation Payload. The messages with an actu-

ation payload from the CC are addressed to the

actuators available in the house. These messages

shows what the objects must perform according

with the data gathered from the environment. The

payload contains information such as the time

stamp of the message, the unique identiﬁer of the

actuator, the name and type of the objects and ﬁ-

nally the command that it must perform.

• Conﬁguration Payload. The messages with a

conﬁguration payload from the CC contain a de-

scription of the triggers and controls that the

house should be able to run autonomously to re-

spond at changing conditions in the environment.

These are messages used to reconﬁgure the home

gateway due to a change in the assistive scenario.

The content foresee a timestamp related to the

message generation and a list of triggers to be in-

stalled (or removed) form the actual conﬁguration

of the home.

3.1.2 Local Applicative Model

The PAss is locally managed through the home gate-

way based on Freedomotic. Freedomotic adopts a

messaging system based entirely on events. Any

change in the environment or interaction with the soft-

ware by users generates an event (click on the graphic

interface, changing a value on an object, etc.). These

are published on deﬁned channel and intercepted by

triggers. In turn, each trigger can be associated with

one or more commands deﬁning a reaction. In other

words, when a sensor communicates any change in

the environment, an event is generated on a speciﬁc

channel, and if the event is consistent with the trig-

ger, then one or more commands can be sent to the

actuator that can run them.

This mechanism allows the creation of rules for

automation in a very simple way. The rules are ex-

pressed using the template ”if THIS then THAT”

where the part THIS corresponds to a trigger, and

THAT is made by one or more commands exe-

cuted in sequence (similar to the popular approach

in https://ifttt.com/). The rules system allows to hide

the implementation details of the triggers and controls

on virtual objects allowing their use for inexperienced

users. They can manage the house to create automa-

tions via the user interface of the system.

The use of local rules is not enough for the PAss

house. Instead, we must use an ”intelligence” located

outside the house that should be able to aggregate data

from multiple sources, including telemedicine plat-

forms. The integration of different data in the home

allows the derivation of conﬁguration and actuation

messages for the house.

3.1.3 Remote Applicative Model

In the remote Cloud Center, the application model of

the integration platform of the house is extended with

an external logic. The data coming from the house are

stored, normalized through big data technologies and

made available for processing with the Bayesian net-

works to enable applications that detect abnormalities

and predict hazardous conditions.

To meet these requirements, the sensor data must

be integrated and interpreted using proper models and

technologies able to deal with uncertainty conditions,

for example some sensors could not sent the measure-

ments due to a technological problem, and we will

work in a non-deterministic but probabilistic setting.

For this we need a tool that models situations involv-

ing uncertainty as, indeed, the Bayesian networks.

To sum up, the system for scenarios recognition

consists of the following elements that communicate

using the same messaging system described above.

AnIntegrationPlatformforPrivateAssistedHouses

Physical sensors (Smart Objects): sensors that de-

tect what happens inside the house and provide the

raw data for the CC applicative model; Virtual sen-

sors: software components that aggregate and ﬁlter

raw data by using mathematical models to avoid er-

roneous readings and to describe measurements as a

sample in a stochastic process; Bayesian networks:

component that recognizes system states and scenar-

ios to reach a high-level goal; Actuators: components

that make decisions and allow the system to react au-

tonomously in the house environment.

In the CC we place all the components needed

to make the management and monitoring of smart

homes (for example by caregivers, medical staff, wel-

fare associations), the management of the conﬁgura-

tion of the PAss integration platform, and the exploita-

tion of the telemedicine platform.

4 A DAY IN THE PAss HOUSE

In order to validate the integration platform differ-

ent use cases has been identiﬁed focusing on differ-

ent levels of care. We assumed a light scenario, an

intermediate scenario and one advanced scenario. In

the following we refer to an intermediate scenario as

a good example of what the house can do to support

the daily routine.

”Mark is a retired person of 65 years old, mar-

ried with no children. He had an ischemic

stroke with paralysis of the right upper and

lower limbs 14 months ago. The relevant

aspects of his health condition can be syn-

thesized as: right hemiparesis with a ﬂex-

ion contracture of the upper limb, aphasia

and dysphagia; high blood pressure; severe

tendinopathy of the subscapularis muscle of

the left shoulder; prostatic hypertrophy. While

there is only a minimal chance of a functional

improvement, there is a real danger of regres-

sion, both in the motor functions and in the

cognitive one. In particular, we want to avoid

the loss in the ability to walk and the worsen-

ing of ﬂexion contracture of the upper limb.

Mark was discharged from the rehabilitation

center three months after the stroke, and then

he continued to rehab in Day Hospital (DH).

Once evaluated his socio-health situation, the

physiatrist and social services have suggested

Mark and his wife to move to a PAss house,

more suited to the needs of the family.”

By showing this use case we will also focus, in

subsections 4.1 and 4.2, on two particular situations

that may occur in the house.

”His day begins at 8 am. After waking up,

he is able to get out of bed due to its inte-

grated automation system and with the help

of his wife. Mark is able to walk short dis-

tances with orthosis and a stick support and

can reach quite easily the bathroom where he

can wash in a bathtub with a rising platform,

always with the help of a caregiver. Moreover,

if the transfer should occur in the absence of

caregivers the presence of a motion sensors

system reveal any falls, ensuring a call to the

rescue.”

In this situation, the system will use the mo-

tion sensor Passive InfraRed (PIR) located inside the

house (usually one for room) to detect the movement.

These sensors trace the movements within the home

environment and, together with the pressure sensors

located on the sittings, we can monitor the activities

carried out identifying abnormalities. As outlined in

the previous sections, these conditions will be moni-

tored through the Bayesian network framework where

the network states use the data incoming from sensors

in order to establish the conditions of normality and

abnormalities.

The normal condition implies that at least one sen-

sor registers the presence of the person through the

recognition of a movement (with the exception of the

person that is not in the house). Terms of inactiv-

ity (for example, TV watching, sleeping) are consid-

ered normal if they involve the use of furniture such

as chairs or bed with a pressure sensor that detects

the use. Abnormality conditions can depend on: (i)

Fall or prolonged inactivity: PIR sensors do not de-

tect movement for a long period of time and/or is not

recognized the use of chairs/bed; and (ii) Malfunction

of the sensors: the absence of data or conﬂicting data

that would result in uncertainty in the assessment of

the scenario.

”Mark can reach the kitchen in an autonomous

way. The morning is the part of the day dedi-

cated to rehabilitation aimed primarily at pre-

venting complications due to hypomobility of

limbs and to increase physical performance.

During this phase, as for the rest of the day, it

is very important to measure vital parameters

through a wearable system of sensors. This is

aimed to monitor at a distance the condition of

hypertension.”

While the environment is monitored, the person

is constantly controlled by wearable sensors that are

able to detect vital signs (heart rate and respiration,

activity level and posture). These information are

involved in the assessment of the condition of nor-

mality/abnormalities through the Bayesian network

ICT4AgeingWell2015-InternationalConferenceonInformationandCommunicationTechnologiesforAgeingWelland

e-Health

framework. If the system detects a situation of in-

activity, the condition of normality can be detected

through the vital parameters measured by the wear-

able sensors. Moreover, the sensors are able to de-

tect the posture (vertical or horizontal) and the accel-

eration along three axes. This is useful to determine

emergency conditions such as the occurrence of a fall

(always considering the data provided by the other

sensors in the house).

”The afternoon is the time for relax and the

presence of a caregiver within the home is

not essential. Mark, using a simpliﬁed inter-

face, can provide for the regulation of temper-

ature and brightness, enabled by the presence

of liquid-screen windows to mitigate sunlight.

Here, the environmental sensors (light meter, tem-

perature/humidity, gas) monitor the conditions inside

the house. The supplied data are used to adjust the

conditions of brightness and temperature in the en-

vironment either automatically, or manually by the

user through an interactive graphical interface. The

data collected from the monitoring system are corre-

lated with information from other sensors in the house

(PIR, pressure) again by applying the Bayesian net-

work framework. For example, if at a certain time of

the day the person is in the living room and the light

meter registers an excessive lighting, the system can

respond automatically activating the darkening of the

liquid-screen windows.

4.1 Fall Recognition

In the unpleasant situation of a fall, we can have

the intervention of several sensors: motion (PIR) and

pressure (chair) sensors as well as wearable sensor

(posture) giving the following information: (1) Move-

ment in the kitchen and chair use, (2) Movement in

the bathroom, (3) No movement in any room, with

speeding down, person in a horizontal position, (4)

Movement in bath, speeding down, person in a hori-

zontal position.

In the above cases, we outlined two different

levels of emergency that can be summarized in a

Bayesian network: (i) The person is still and may

have lost consciousness; (ii) The person moves (de-

tected by the PIR in the bathroom). The condition

of horizontal position and the detection of a down-

ward acceleration are considered fundamental situa-

tions because they are considered an abnormality in-

side the bathroom.

4.2 Adjustments of Environment

Some scenarios can involve environmental monitor-

ing sensors (temperature/humidity, light meter), mo-

tion sensors (PIR, pressure), an interactive panel, a

graphical interface and the liquid-screen window. For

example, in a bedroom the data provided can be the

following: (i) Movement in the bedroom, not enough

room lighting (the system responds by turning on the

light in the room); (ii) Movement in the kitchen, tem-

perature below the average (the system performs a

temperature regulation modulating heating); (iii) The

operation of the window is performed manually by

the user to answer a particular need.

5 RELATED WORK

In the context of Internet of Things and Ambient As-

sisted Living much interest has been placed into the

home environment to address issues related to med-

ical care and for the comfort and welfare of the in-

habitants. Several projects and works have been de-

veloped, each with different characteristics and target.

We report some works that are most close to our.

In (Kaldeli et al., 2013) the authors report the ef-

fort made in the European project SM4ALL that aims

to build a smart home able to exhibit complex func-

tionalities build over a set of services offered by real

objects and the environment. The system is charac-

terized by a composition of services made using plan-

ning techniques that deals with devices that evolve

over time exposing new functionalities or disappear-

ing from the environment. In such high dynamic con-

text declarative goals are stated by the user (or in-

ferred by rules) and then a plan of actions is made

by the planner and realized by the pervasive layer of

object in the house.

The MavHome Smart Home research project

equips the house with the ability to make decisions

based on predicted activities looking for pattern of de-

vice utilization and movement in the house (Rao and

Cook, 2004). Actions of the inhabitants are modeled

as states in Markov models to provide automation and

adaption to the inhabitant’s needs.

In (Pellegrino et al., 2006) an architecture of a

home automation gateway is presented. This ap-

proach supports the integration of heterogeneous de-

vices and uses an enhanced run-time engine to gener-

ate events at run-time basing either on events coming

from the house or by inferred rules to prevent annoy-

ing or dangerous situations.

Some efforts are also made by the Am-

bient Assisted Living Joint Programme

AnIntegrationPlatformforPrivateAssistedHouses

(http://www.aal-europe.eu/. The problem of

monitoring vital signs is faced by EMOTION-

AAL (http://www.emotionalaal.eu/) and H@H

(http://www.health-at-home.eu) projects that aims to

build integrated platforms for collecting data from a

variety of bio-sensors for the permanent monitoring

of the state of health of the users. Fall detection

is also investigated, among others we can mention

CARE (http://www.care-aal.eu) and ROSETTA

(http://www.aal-rosetta.eu/). They provide a way to

monitor, analyze and interpret the behavior of the el-

derly at home (such as falls or loss of consciousness),

and automatically generate an emergency call.

6 CONCLUSION

This work is the results of a public ﬁnanced action for

the development and implementation of an integra-

tion platform for Ambient Assisted Living to moni-

tor activities of daily living and to detect any abnor-

mal behavior that may represent a danger, or high-

light symptoms of some incipient disease. According

to the need of elderly people, Private Assisted House

is also an enabling technology for the development of

a novel care model that considers the changing needs

of users using an highly conﬁgurable integration plat-

form aimed to support their daily life.

Our future works will go in several directions:

(i) the prototype and care model will be evaluated

in rehabilitation institutions of Marche Region (Italy)

to allow an experimental evidence of the working

system and to provide feedback for further techni-

cal developments; (ii) we want to extended out plat-

form over time with new components to solve speciﬁc

problems or address the needs of a speciﬁc category

of people (or disabilities, or disease) integrating also

the development of speciﬁc tools and smart object;

(iii) we intend to further improve the application of

the Bayesian networks in order to implement a sce-

nario where the system changes over time, by learn-

ing or by changing conditions of normality and ab-

normality based on habits of the subject who lives in

the PAss house; (iv) we want to work on a care model

suitable for Marche Region institutions to bring AAL

to the user’s house.

In this respect, this work represent the core of an

infrastructure that will be used in real scenarios and a

special attention will be devoted to the result obtained

not only form the technological perspective but also

from the quality of life perception form the end-users.

ACKNOWLEDGEMENTS

The project PAss ”Private Assisted House”

(www.projectpass.eu) is co-funded by the Marche

Region administration, under the action ”Smart

Home for Active and Healthy Aging”, so we thanks

to Marche Region and all the partners of the project.

REFERENCES

Chang, F., Dean, J., Ghemawat, S., Hsieh, W. C., Wallach,

D. A., Burrows, M., Chandra, T., Fikes, A., and Gru-

ber, R. E. (2006). Bigtable: A distributed storage sys-

tem for structured data. In 7th Conference on USENIX

Symposium on Operating Systems Design and Imple-

mentation, volume 7, pages 205–218.

Dean, J. and Ghemawat, S. (2004). Mapreduce: simpli-

ﬁed data processing on large clusters. In OSDI ’04.

USENIX Association.

Freedomotic web site (2014).

http://www.freedomotic.com/.

Gubbi, J., Buyya, R., Marusic, S., and Palaniswami, M.

(2013). Internet of things (iot): A vision, architectural

elements, and future directions. Future Generation

Computer Systems, 29(7):1645 – 1660.

Kaldeli, E., Warriach, E. U., Lazovik, A., and Aiello, M.

(2013). Coordinating the web of services for a smart

home. ACM Trans. Web, 7(2):10:1–10:40.

Kortuem, G., Kawsar, F., Fitton, D., and Sundramoorthy, V.

(2010). Smart objects as building blocks for the inter-

net of things. Internet Computing, IEEE, 14(1):44–51.

Mace, R. L., Hardie, G. J., and Place, J. P. (1990). Acces-

sible environments: Toward universal design. Center

for Accessible Housing, North Carolina University.

Mqtt web site (2014). http://mqtt.org/.

Nehmer, J., Becker, M., Karshmer, A., and Lamm, R.

(2006). Living assistance systems: An ambient in-

telligence approach. In Proceedings of the 28th Inter-

national Conference on Software Engineering, ICSE

’06, pages 43–50, New York, NY, USA. ACM.

Pearl, J. (1988). Probabilistic reasoning in intelligent sys-

tems: networks of plausible inference. Morgan Kauf-

mann.

Pellegrino, P., Bonino, D., and Corno, F. (2006). Domotic

house gateway. In Proceedings of the 2006 ACM Sym-

posium on Applied Computing, SAC ’06, pages 1915–

1920, New York, NY, USA. ACM.

Rao, S. P. and Cook, D. J. (2004). Predicting inhabitant

action using action and task models with application

to smart homes. International Journal on Artiﬁcial

Intelligence Tools, 13:81–100.

Rossi, L., Belli, A., De Santis, A., Diamantini, C., Frontoni,

E., Gambi, E., Palma, L., Pernini, L., Pierleoni, P.,

Potena, D., Raffaeli, L., Spinsante, S., Zingaretti, P.,

Cacciagrano, D., Corradini, F., Culmone, R., De An-

gelis, F., Merelli, E., and Re, B. (2014). Interoper-

ability issues among smart home technological frame-

works. In IEEE/ASME, pages 1–7.

ICT4AgeingWell2015-InternationalConferenceonInformationandCommunicationTechnologiesforAgeingWelland

e-Health