New Software Solutions for Low-power Management of Green Smart

Homes

∗

Aymen Jaouadi

1,2

, Olfa Mosbahi

, Mohamed Khalgui

and Asma Sakri

Cynapsys Company, France-Germany

FST, University of Tunis, El Manar, Tunisia

LISI Laboratory, INSAT Institute, University of Carthage, Tunis, Tunisia

Keywords:

Microcontroller, STM32F4, Reconﬁguration, Smart Home, Smart Grid, Software Agent, Modeling and

Veriﬁcation, Simulation.

Abstract:

The research paper deals with new Green Smart Homes which offer original services such as the optimal power

consumption, peak management, and home power selling while assuming an available home green energy.

We propose a Master-Slave based architecture following the well-known industrial technology STM32F4. A

microcontroller Slave Agent is proposed for each selected Home Device to control its local consumption, and

a unique microcontroller Master Agent is proposed to control the whole architecture. The goal is to optimize

the use of green energy, to minimize the consumption costs by exploiting the offers from providers and also

the peak times. We model these services by using the model checker UPPAAL, and propose UML design

diagrams for this architecture. A visual simulator of this STM32-based architecture is developed and applied

to a case study proposed by Cynapsys.

1 INTRODUCTION

Home automation consists in the management of our

equipments based on new technologies and also in

the proper use of information. If it is properly used,

the home automation can make real energy savings

and be part of a sustainable development of the future

electricity networks. Home automation presents itself

as an alternative to reduce energy costs, increases in-

telligence at homes by the mean of sensors and actu-

ators, as well as maintaining the power grid by the

balance of supply and demand. To make life eas-

ier, Home Automation is an obvious choice, it allows

to design a green grid, which is based on renewable

energy and to encourage innovation through the lat-

est discoveries in the ﬁeld of high tech. In recent

years, several studies about the automation were con-

ducted; the most important proposes an approach that

is based on the diversiﬁcation and the decentralization

of energy production sources and the development of

∗

This research work is a collaboration between LISI

Laboratory (INSAT Institute) at University of Carthage and

the Company Cynapsys in Tunisia. We thank Ing. Haythem

Tebourbi Technical Director and Ing. Souhail Kchaou Di-

rector of Research and Development for long fruitful dis-

cussions and stable supports.

a platform that enables the exchange and communi-

cation between different types of interconnected net-

works (Torbensen, 2008). Inspired by (Abras et al.,

2008) home automation systems can be combined

with Multi-Agent Systems for the home energy man-

agement and the adaptation of the consumption to the

available green energy resources as well as the de-

velopment of new algorithms for emergency and an-

ticipation mechanisms. Among the used communi-

cation solutions at home automation systems, Inter-

net, ZigBee or Bluetooth have been proposed to con-

trol households from an remote location by using vi-

sual interfaces and web applications. (A.Vichare and

Verma, 2012)(Gill et al., 2009)(R.Piyare and M.Tazil,

2011). we are interested in this research in optimal

home electricity management through the creation of

new modern Home Automation Services. We propose

an original reconﬁgurable home automation system

Model which allows the dynamic change of the daily

behavior at run time according to user requirements

and energy constraints. The reconﬁgurable model of

our approach consists of the execution and the con-

trol of three runtime services: a Power Consump-

tion Management Service which is used to manage

and control the power consumption of equipments at

home. It adapts the energy demands to production

375

Jaouadi A., Mosbahi O., Khalgui M. and Sakri A..

New Software Solutions for Low-power Management of Green Smart Homes.

DOI: 10.5220/0004976003750385

In Proceedings of the 3rd International Conference on Smart Grids and Green IT Systems (IEEHSC-2014), pages 375-385

ISBN: 978-989-758-025-3

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

systems through an action plan predeﬁned by the user.

The second service consists of the Management of

the green low power distribution which is based on

the decentralization of production through the use of

renewable sources in low-voltage networks. This en-

ergy is stored in batteries and used according to a spe-

ciﬁc management and consumption policy. This ser-

vice also allows the user to manage sales promotional

offers, it may as well buy energy during discount pe-

riods, store it and use it when prices are inﬂated. Fi-

nally we propose the third peak management service

which allows to manage high consumption periods

by deactivating directly the equipments or delaying

there execution. This service also allows to compen-

sate the difference between the production and the

consumption thanks to use of green energy that we

assume stored in batteries. These original services

meet exactly the different requirements of the modern

and smart electricity grid. We propose a Multi-Agent

based architecture to offer these services. It is com-

posed of: (a) City Information Agents to gather the

useful information for the distribution and the use of

electrical energy, (b) City Master Agents that control

Home Master Agents which supervise Home Slave

Agents to be responsible for the operation of the elec-

trical equipments at homes, and ﬁnally (c) Home Stor-

age Agents to be deployed for the monitoring and the

exploitation of renewable energy produced and stored

in batteries at home. An original Master/Slave archi-

tecture is hierarchically proposed in the current pa-

per to manage Information, Storage and the proper

use of green electricity. This architecture is guided

by a communication protocol that manages the ex-

changed messages between the different agents in or-

der to perform the services discussed above. This

optimal protocol allows the management of renew-

able energy, management of consumption peaks and

management of electricity consumption. We also pro-

pose the formalization of these services and the de-

sign of the software with UML diagrams. We verify

the whole architecture by proposing timed automata

models, and apply UPPAAL for model checking(Alur

and Dill, 1994),(Bengtsson et al., 1996). We aim to

verify functional and temporal constraints since our

system is strict and does not tolerate faults and mis-

management of used data and information. The study

of our approach leads us to develop a simulation tool

to be named X-SH which is an original product for

the power management in smart homes. The paper’s

contribution is applied to a real case study provided

by Cynapsys in order to discuss its advantages. The

rest of this paper is organized as follows, the next

section presents the state of the art of Smart Homes.

Section 3 presents the Case Study of Cynapsys that

will be assumed as a running example in this paper.

Section 4 proposes the Multi-Agent Master/Slave ar-

chitecture followed by a formalization of the problem

and the proposed services, the UML design and the

veriﬁcation of timed automata models, The Simula-

tion tool X-SH. Finally section 5 concludes this re-

search works.

2 BACKGROUND

We present in this section an overview on Home Au-

tomation Systems, System Reconﬁguration and Mas-

ter/Slave Agent based architecture.

2.1 Home Automation Systems

Nowadays, the research works in the ﬁeld of home au-

tomation systems have a unique direction to look for

luxury, comfort and the mixture between daily tasks

and new technologies. The authors in (AlShu’eili

et al., 2011) propose a new approach for voice recog-

nition based wireless home automation system to con-

trol all lights and electrical appliances at home or of-

ﬁces by using voice commands, they propose a ver-

iﬁcation test based on the voice recognition. How-

ever, this approach is weak against identity theft and

imitating the voice of the house’s owner, so anyone

can take control of the home. According to (Nunes,

2010), an architecture for a home automation system

is given, which has a distributed nature, very mod-

ular and can easily be expanded in size and func-

tionality. The authors present two types of constitu-

tive modules interconnected through a network. The

proposed approach consists of Supervision Module

(SM) and a Control Module (CM) to be intercon-

nected by a Communication Network, a simple sys-

tem with an Action/Reaction mechanism. In (Debono

and Abela, 2012), the authors present an implemen-

tation of a home automation system through a cen-

tral FPGA controller as a simple solution whereby the

user control devices by employing a central ﬁeld pro-

grammable gate array (FPGA) controller to which the

device and sensor are interfaced. The control is es-

tablished by using a communication with the FPGA

from a mobile phone through its interface. A design

of a networked monitoring System for home automa-

tion is presented in (Song et al., 2007). The system

consists of a base station, a home server, wireless sen-

sor nodes and smart user terminals such as PC and

PDA. A ZigBee-Based Home Automation System is

developed in (Gill et al., 2009). the authors present

a ﬂexible and low cost home automation infrastruc-

ture. The home’s low data rate, control and monitor-

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

376

ing needs are catered by using ZigBee. the authors

in (Ha, 2009), try to present a dynamic integration

of ZigBee home networks into home gateway by us-

ing OSGi service registry. The proposed architecture

is divided into two layers of ZigBee home networks;

physical and logical interface. A realization of home

remote control network based on ZigBee is described

in(Shunyang et al., 2007); the authors present the Zig-

Bee as an emerging wireless communication technol-

ogy with low cost and low power characteristics.

2.2 Reconﬁgurable Embedded Systems

We assume that an embedded system is reconﬁg-

urable if it changes its software or hardware behav-

ior at run-time according to user requirements. The

software reconﬁguration is any operation allowing the

addition, removal or update of software tasks that im-

plement the system to encode corresponding func-

tions. The software reconﬁguration is assumed to be

any operation allowing the addition, removal or up-

date of hardware components according to user re-

quirements. An addition or removal can be of mem-

ory, of data-event inputs-outputs, or of a new net-

work for communication. The update of hardware

components can be the modiﬁcation of the proces-

sor speed. We are interested in this paper in soft-

ware reconﬁgurations. We distinguish two reconﬁg-

uration policies: static and dynamic reconﬁgurations

(Wang et al., 2010). Static reconﬁgurations are ap-

plied off-line to implement changes before the cold

start of the system (Angelov et al., 2005), whereas

dynamic reconﬁgurations are dynamically applied at

run-time.(Seokcheon, 2010)(Hu et al., 2011) Two

cases exist in the latter: manual reconﬁgurations ap-

plied by users(Rooker et al., 2007) and automatic

reconﬁgurations to be applied by intelligent agents

(Seokcheon, 2010)(Khalgui et al., 2011).

2.3 Master/Slave Multi-agent

Architecture

According to the authors in (Megherbi and Madera,

2010), the Master-Slave architecture is described as

the most popular and widely used architecture in the

distributed systems. Master software nodes assign

each slave node for a speciﬁc amount of work, and

once the slave has completed its task, it reports the

results back to the master. In (Nwana et al., 1996),

the authors propose a coordination model in software

agent systems, based on Master-Slave architecture,

the master agent plans and distributes fragments of

the plan to the slaves. The slaves may or may not

communicate among themselves, but must ultimately

report their results to the master agent.

2.4 Contribution: New Services for

Smart Homes

According to the study of these research works pro-

posed in Smart Grids, Home Automation Systems,

Multi Agent Systems, and Reconﬁgurable Embedded

Systems, we ﬁnd that the researches carried out: do

not respond exactly to the requirements of the mod-

ern world, do not consider the exponential growth of

the energy consumption, and do not follow the new

technical and technological progress. These works

are simple, wobbly, and too generalizing. Indeed they

lacked precision and intelligence. We propose in the

current paper a new Master/Slave Multi-Agent archi-

tecture in order to: (a) decentralize the control, (b)

guarantee the determinism, (c)and reduce the number

of messages circulating in the network. This original

architecture allows: (a) the management of the energy

consumption, (b) the management of peak period, (c)

the use of renewable green energy to be stored in bat-

teries, (d) and the management of the promotional of-

fers from providers when the prices are down.

3 CASE STUDY: CYNAPSYS

HOME AUTOMATION SYSTEM

We present in this section the Cynapsys Smart Home

(to be denoted by CSH) which was maintained as

the reference case study of our company in the cur-

rent project. We assume a house to be composed

of six rooms and a garden. Each home area is sup-

posed to be controlled by a STM32 F4-based elec-

tronic equipment for the optimal daily energy con-

sumption (STMicroelectronics, 2013). CSH ensures

serenity and control of ambient lights, appliances and

access. The Proposed CSH provides automated blind

controls to open or close, and to be based on times

or light and heat levels. An integrated cooling and

heating system to optimize the energy use through

pre-determinedscheduling or temperaturecontrols, as

well as the management of the smart meter to con-

trol electricity consumption in real-time and the pro-

motional offer of the energy providers. We propose

a control irrigation system and pool pump operation

management and ﬁnally the integration of new green

renewable energy production sources to be composed

mainly by wind and solar panels that we store in home

batteries. The CSH shows many interests: (a) the

billing service, (b) the conservation of energy, (c) the

NewSoftwareSolutionsforLow-powerManagementofGreenSmartHomes

377

peak management, (d) and the use of renewable en-

ergy. Figure 1, provides a description of CSH that we

assume as a running example in this paper.

Figure 1: CSH case study.

We propose the characteristic of the CSH equip-

ments in the table below with real values about the

energy consumption, the frequency of use and the pri-

ority of each device.

Table 1: Electrical Equipment Characteristics.

Equipment P UF CA Pr

Combined Fridge 100 W Continuously 875 2

Air Conditioning 150 W 5 H/day 450 2

Washing machine 940 W/C 3 cycles/week 135 1

Microwave 500 W 1.5 H/Week 36 1

Boiler 2000 W 80L/day 400 1

Heating 60 W Continuously 345 2

Pool Pump 1500 W 5H/day 750 1

Where:

P: is the equipment power (watt)

UF: is the equipment usage frequency

CA: is the equipment Calculated amount (KWH)

Pr: is the equipment Priority.

The data presented above will be used in the imple-

mentation of our proposed new services such as the

management of peak demand or help the user to ra-

tionalize its energy consumption and pay less. In this

section, we propose a novel idea which is based on

the classiﬁcation of household tasks into two broad

categories, the ﬁrst category contains the permanent

tasks, i.e the daily tasks that have the highest prior-

ity such as: radiators and heater. The second cat-

egory contains temporary tasks that can be delayed,

such as television, washing machine, dishwasher and

iron. We will assume that we have three tariff zones

in a day as described in Table II: (a) from midnight to

8am, (b) from 8am to 16pm is a middle zone and (c)

from 16 pm until midnight. The ﬁrst zone is used for

the normal consumption and we attribute coefﬁcient

2 to this period. The second one is characterised by a

consumption peak and has the high coefﬁcient equal

to 4. The last one is a medium zone with consumption

coefﬁcient equal to 3.

Table 2: Pricing zones.

Normal Peak Medium

Zone Zone Zone

Time Slot 00.01 to 8.01am to 16.01pm to

Time Slot 8am 16pm 00.00

Coefﬁcient 2 4 3

Price Mill/KWH 130 260 195

In the following, we will present the advantages of

using three tariff zones for both continuousand partial

time tasks.

- For a continuous task, for example refrigerator:

– With a unique tariff (peak zone): The annual

consumption in Tunisian Dinar is equal to:

875∗ 0.260 = 227.5 TND/Year. (1)

– With three tariff zones: The annual consump-

tion in Tunisian Dinar is equal to:

(875∗ (0.130+ 0.260+0.195) ÷ 3) = 170.6 TND/Year.

(2)

- For a partial time task, for example the pool pump:

– With a unique tariff (peak zone): The annual

consumption in Tunisian Dinar is equal to:

(750∗ 0.260) = 195 TND/Year. (3)

– With Normal tariff zone: The annual consump-

tion in Tunisian Dinar is equal to:

(750∗ 0.130) = 97.5 TND/Year. (4)

In the ﬁrst case, we notice a gain in the bill equal

to (227.5− 170.6) TND. In the second one, the user

choices the cheapest pricing zone and can earn the

half of the price (195− 97.5) TND.

To test the peak management service, the elec-

tricity daily consumption estimation during a summer

day is as follows (using Table I):

(2.4+ 0.75+0.45+ 0.11+ 1.09+7.5) = 12.3 KWH/Day.

(5)

The available energy during peak consumption period

is equal to 2 KWH/Day. So in this case, the requested

value is equal to 10.3 KWH(12.3− 2). We propose

to deactivate the less priority equipments during peak

periods. So we will only keep the refrigerator and air

conditioning and the daily consumption will be equal

to:

(2.4+ 0.75) = 3.15KWH/Day. (6)

Compared to the provided value, we still need

1.2 KWH to meet the daily electricity lack. We as-

sume that we have domestic batteries to store energy

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

378

produced by renewable sources. The used batteries

are divided into two parts. The ﬁrst part uses 70% of

the available capacity for optimizing the production.

The second part uses 30% to solve the peak consump-

tion problems. For example, if the battery available

amount of energy is equal to 4KWH, we have to use

only 30% of this amount, which is equal to 1.2KWH

After using our idea, we we

re able to avoid a peak consumption by delaying

the execution of some lower priority equipment and

compensate the remaining value through the use of

renewable energy available in the battery.

4 MULTI-AGENT BASED

SOLUTION

In this section, we propose a Multi-Agent based Ar-

chitecture, for our CSH to control and manage the do-

mestic energy consumption. Our architecture consists

of a one Home Master Agent controlling a number of

Slave Agents corresponding to their equipments. We

have also Storage Agents functioning together with

the Green Agents for the production and the storage

of the green renewable energy.

Figure 2: Multi-agent based architecture.

4.1 Agents Description

In this section we will introduce the various agents

used in our approach. We have four kinds of agents:

1. Home Master Agent. It controls and optimizes

energy consumption and encourages the use of the

green renewable sources. It coordinates between

all the Slave Agents controlling their correspond-

ing equipments.

2. Home Slave Agent. They control the elec-

trical equipments by stopping and resuming

their functioning through a mechanism of Sen-

sors/Effectors,

3. Green Agent. They control the production of the

renewable energy sources,

4. Storage Agent. They control storing in batteries

the energy produced by green renewable sources,

to be used in the case of peak consumption.

4.2 Agents and Services Formalization

In this section we propose a formalization of the

agents and the original services used in the contribu-

tion.

4.2.1 Formalization of Agents

Let

• Sys

CSH

= { MA,ξ

,ξ

STA

} be the Cynapsys

Smart Home Multi-Agent based System com-

posed of a Master Agent MA, N Slave Agents

SA, M Green Agents GA and ﬁnally K Storage

Agents. Let:

• ξ

= {SA

,...,SA

} be the set of Slave Agents

for the control of rooms,

• ξ

= {GA

,...,GA

} be the set of Green Agents,

• ξ

STA

= {STA

,...,STA

} be the set of Storage

Agents,

• Sys

= {e

,...,e

} be the set of electrical

equipments.

4.2.2 Agent’s Parameters

Let MA = {ID, EC,ASL,NLV} be the Master Agent

list of parameters, where:

• ID: the Identiﬁer of the agent composed of a

String,

• EC: the daily estimated energy consumption,

• ASL: the Available Stored Load energy,

• NLV: the Needed Load Value used in the case of

peak consumption and daily production manage-

ment.

Let e

∈ Sys

where e

has:

• ID: be the Identiﬁer of the electrical equipment,

• EC(e

): be the daily estimated consumption of the

electrical equipment,

Let sa

∈ ξ

where sa

has an ID, an NLV, and and a

set of equipment Sys

to control.

• ID: be the Identiﬁer of the Slave Agent,

NewSoftwareSolutionsforLow-powerManagementofGreenSmartHomes

379

• NLV(e

): be the Slave Agent Needed Load Value

corresponding to the equipment e

Let ga

∈ ξ

where ga

has an ID and a DPV

• ID: the Identiﬁer of the Green Agent,

• DPV: the Daily Production Value.

Let sta

∈ ξ

STA

where sta

has an ID and a STV

• ID: the Identiﬁer of the Storage Agent,

• STV: the Stored Value.

4.2.3 Formalization of Services

The intelligence in power systems spreads starting

houses, through intelligent cities to arrive at a smart

grid.

Let:

• P

be the equipment power amount,

• P

be the general available energy.

Our system should always satisfy the following equa-

tion:

∑

i=1

ECe

< P

(7)

let:

• Pr

be the total price of the energy consumption,

• Pr

be the price at the instant i,

• C

be the consumption at the moment i,

• Pu

be the unit price of the KWH at the moment i,

• Pr

= C

∗ Pu

• Tr

be the Consumption Threshold.

Our approach must satisfy the system represented by

the equation below, which consist of using the three

tariff zones, i ∈ [0, 1,2] where zero corresponds to the

ﬁrst tariff zone, one for the second one and two for

the third one:

∑

i=0

(8)

The problem is to ﬁnd the right formulas of energy

consumption during the appropriate periods to mini-

mize the total cost.

The equation is therefore to establish an optimization

relationship having the form:

• MinPr

• Pr

< Tr

Let:

• EC be the estimated daily consumption of a

house,

• ECe

be the estimated consumption of a single de-

vice e

per day.

The estimated daily consumption is given by the fol-

lowing equation:

EC = ECe

+ ECe

+ ... + ECe

∑

i=1

ECe

(9)

Let:

• Pv be the Peak Value provided by the energy pro-

ducers,

• STV be the available capacity in the battery.

In the case of a peak detected by the energy pro-

ducer, the power source become the battery in case of

presence of sufﬁcient load. In our approach, we have

30 % of the capacity of our battery remains untouch-

able, this part is dedicated to solving the problem of

peak consumption as described in the equation below:

• Pv = 0.3 ∗ STV

• STV

be the stored value of the Storage Agent i.

4.3 Service Protocols

In this section, we propose a Service Protocol to co-

ordinate all used agents in the implementation of the

different proposed services.

Algorithm 1: Green Energy and Storage Device

Management.

foreach ga

∈ ξ

STV

∑

i=1

DPV

end

// The Sum of the Daily produced Value

provided from different green renewable

sources represents the Stored Value used in the

Peak Management Service.

foreach sta

∈ ξ

STA

STV =

∑

i=1

STV

end

if Date == Date

Peak then

Pv = STV ∗ 30 ÷ 100;

//The Peak Value represents 30% of the

available capacity on the Battery

end

The next algorithm deals with the management of

the interaction and the communication between the

used agents used in order to handles the power con-

sumption optimization.

4.4 Timed Automata Models

In this paper, we propose a global approach for the

design of adaptive reactive systems, i.e. systems that

dynamically adapt their architecture depending on the

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

380

Algorithm 2: Home Master Agent and Home Slave

Agents Energy Management.

Init

Date <> Date

Peak

State = Enabled

while Date = Date

Peak do

foreach sa

∈ ξ

if Priority == 1 then

State == Disabled

end

if NLV <= Pv then

MA Sends a giveSolution(MA,STA)

else

foreach sa

∈ ξ

State = Disabled

end

if Event == PromotionalOf fer then

MA Sends Store(MA, STA)

else

foreach sa

∈ ξ

if (Priority == 1) And

(Zone <> NormalZone) then

State = Disabled

end

if Pr

>= Tr

then

foreach sa

∈ ξ

State = Disabled

end

State = Enabled;

end

context. We use the timed automata formalism for

the design of the agent’s behavior to check and verify

functional and temporal properties of our inter-agent

communication protocol. This allows their evaluation

regarding logical correctness and timeliness, thanks

to model-checking and simulation techniques. Ac-

cording to (Palshikar, 2004), model checking is the

most successful approach that’s emerged for verify-

ing requirements. The idea is that by ensuring that

the model satisﬁes enough system properties, we in-

crease our conﬁdence in the correctness. As follows

we present a Timed Automata Model for the example

of agents that we propose in our approach. We model

the Home Master Agent with its two main important

services: Distribution and Peak Management which

has been described above. The Home Master Agent is

responsible of the whole house control by supervising

a set of Home Slave Agents which in their turns are

responsible of the home’s electrical equipments. Fig-

ure 3 shows the timed automata model of the Home

Master Agent.

Peak_Management

Normal_Mode

Promotion_Management

Ditribution_Management

clpromo:=0

Control_Mode

clnr:=0 Nr:=1

clnr:=0 Nr:=0

Idle

needed_Value[ID]!

estimated_Value[ID]!

evt==date_promo

produced_Value<estimated_Value

date=date_peak

needed_Value<produced_Value

clctr:=0 ctr==0

clnr:=0 Nr:=1

date=date_peak

consumed[ID]>treshold[ID]

clpeak:=0

needed_Value>produced_Value

date!=date_peak

consumed[ID]<treshold[ID]

date>date_promo

clnr:=0 Nr:=1

clctr:=0 ctr:=1

clpeak:=0

needed_Value>produced_Value

clctr:=0 ctr:=1

Consumed[ID]=Treshold[ID]

clpeak:=0

needed_Value<produced_Value

cldist:=0

clpeak:=0

date>date_peak

clctr:=0 ctr:=1

Figure 3: Home Master Agent Timed Automata.

This automata consists of different states related

to the different operations of the Home Master Agent.

The default mode is the normal mode, however, the

transition to the control mode is activated by: (a)

the arrival of preemptive events, (b) functional con-

straints related to exceeding consumption threshold,

(c)the arrival of a peak load, (d)time constraints re-

lated to power management across shutdown, (e) and

activation of equipments to consume intelligently and

during speciﬁc periods. Figure 4, shows the transac-

tion of distribution management, which is an impor-

tant service in our theoretical approach. The decen-

tralization of the distribution in this work is a major

asset for ensuring optimization, maintaining the net-

work and the integration of renewable sources for the

energy generation. The reactive system upon detec-

tion of failure or the arrival of a new promotional offer

from the provider. In this case, we verify the amount

of renewableenergy to be available in the battery. Fol-

lowing the application of the formulas given above,

our system decides if the new source of energy an be

used or not. Our reactive system will be efﬁcient in

this case while respecting imposed time and ecologi-

cal constraints.

init()

Provider_Source

Battery_Source

isFault()

isFault() init()

isPromo()

battery_Amount<estimated_Value

battery_Amount>=estimated_Value

Battery_Check

battery_Amount>=estimated_Value

Idle

clbattery:=0

isFault

init()

Figure 4: Distribution Management Service Timed Au-

tomata.

NewSoftwareSolutionsforLow-powerManagementofGreenSmartHomes

381

The proposed approach allows the management of

peaks by using useful information to be sent from the

provider. Figure 5 describes the mechanism of energy

management during peak periods. Our goals: (a) the

minimisation of the energy consumption during these

periods, (b) take into account environmental impacts

and cost of energy productions, (c) optimisation of the

overall energy generation plan by anticipating con-

sumer demands while ensuring comfort, (d) taking

into account resources and environmental constraints,

(e) and better control of consumption and local distri-

bution. The management of high consumption peri-

ods is based on the use of compensation for energy re-

quests through house storage devices to be connected

essentially to green production sources. The system

is converted to a control mode after the arrival of a

peak load event. The source of production becomes a

local battery after the checking of its capacity. This is

done with a speciﬁc management policy with efﬁcient

calculation methods to guarantee green and low cost

energy for users.

Battery_CheckAlertIdle

isPeak()

Battery_Source

Control_Mode

isPeak()

Battery_Amount>=Estimated_Valueavailable_Value<Estimated_Value

date>date_peak

Peak_Battery_Amount<Estimated_Value

clpeak:=0

isPeak()

Figure 5: Peak Management Service Timed Automata.

4.5 Implementation

4.5.1 UML Modeling

We propose a UML model to implement the software

version of our home system. Figure 6, shows the

class diagram ”Home Automation System” (abbrevi-

ated by HAS) that represents the whole Multi-Agent

architecture. This class manages all interactions in

the system. It is related to the provider interface class

which allows interaction with energy producers, and

the user interface class which allows manual accesses

of users into the system. An Agent class is proposed

to represent both Home Slave Agents and Home Mas-

ter Agent through an inheritance relationship. Figure

6 shows the whole class diagram.

Figure 6: Home Automation System class Diagram.

4.5.2 X-SH Simulation Tool

The simulation is mandatory for the development and

deployment of our system model. This model repre-

sents the main characteristics of the selected physi-

cal, abstract system or process. Simulation is used in

many contexts, such as simulation of technology for

performance optimization, safety engineering, test-

ing, training and education. In this section, we present

a tool to be called X-SH that we developed at Cynap-

sys for the electrical energy management. This tool

simulates the services that we describe aboveby using

real-time data collection.It reacts to various changes

of the electrical network. This tool can be used by

both users and producers for the estimation of the

daily energy consumption in order to optimize the use

of green sources, manage peak periods, manage offers

from producers. After the authentication interface to

be used for a required security, the main interface ap-

pears to include the various offered services that we

describe above: (a) Equipments Management service,

(b) Consumption Management service, (c) Consump-

tion Threshold service, (d) and Sale Management ser-

vice (Figure 7).

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

382

Figure 7: Main Interface Energy Management Tool.

In Figure 8, we show the equipments management

interface. The households are controlled by the Home

Slave Agents. The X-SH Simulation tool allows the

addition of new equipments, removing or changing

priorities according to periods of use and operation

modes.

Figure 8: Equipment Management Interface.

Figure 9, shows the consumption management

service which is provided by our simulation tool X-

SH. The interface informs the users about the values

of the estimated consumption for a period and the ac-

tual daily consumption.

Figure 9: Consumption Management Interface.

In Figure 10, we show the consumption Thresh-

old management Interface. This service allows to

avoid the energy waste or loss as ﬁxed by users. In

the case of exceeding Threshold, we are directly in-

formed about the passage of the home to the control

consumption mode.

Figure 10: Set Threshold Consumption Interface.

The last service proposed by our consumption

management tool X-SH consists of the management

of the provider promotional offers (Figure 11). In the

case of lower prices, the user can change its produc-

tion source from the battery to the provider. We con-

sume less in this case and can help to preserve energy

that we store in the batteries for future uses, manage-

ment electric failures, and management of peak peri-

ods.

Figure 11: Sale Management Service Interface.

4.5.3 Contribution Analysis

The testing of our simulation tool at Cynapsys helped

us to better assess the gain of the proposed approach.

Thanks to this contribution, the management of the

power consumption shows a remarkable gain that

costs 50 TND = 227 TND - 170 TND before the ap-

plication of our approach and after (1 Euro = 2.2 TND

in January 2014). This gain is very important for any

Tunisian family in the medium society class. (Figure

12).

NewSoftwareSolutionsforLow-powerManagementofGreenSmartHomes

383

0 2 4 6 8 10 12

100

150

200

250

month

Dinars

Before contribution

After contribution

Figure 12: Comparison between the energy consumption

before (Red) and after (Green) the contribution.

We propose also in Figure 13, the test of the

thresholds consumption service that we did at Cynap-

sys. We note that during the same period, the user can

consume energy intelligently by considering the pri-

ority of each equipment and the production of renew-

able energy from batteries. The consumption should

not exceed the ﬁxed amount.

0 5 10 15 20 25 30

100

150

200

250

300

day

Dinars

Before contribution

After contribution

Figure 13: Comparison between the energy consumption

without threshold (Red) and with threshold (Green).

Finally we propose in Figure 14 a histogram show-

ing the use of renewable energy in our approach. We

choose a summer day, which is characterized by a

high consumption of electricity and an alert of peak

from the provider. We applied an optimal manage-

ment of the energy with the participation of the re-

newable sources that costs 38.09%.

Figure 14: Green renewable energy participation.

5 CONCLUSIONS

This paper deals with a new Multi-Agent architec-

ture for Smart Homes, which propose, new services

for users such as: (a) Peak Management, (b) Promo-

tion Offers Management, (c) Consumption Manage-

ment. We offer a Home Master Agent that controls the

whole architecture, interacts with the energyprovider,

coordinates with home users in order to optimise the

green energy that we assume available at home. A

Home Slave Agent is proposed for the local control of

each equipment. We propose a communication proto-

col between these agents to support all the services

that we propose above. Since the architecture that we

propose is real-time and based on concurrent opera-

tions, we propose timed-automata models for these

agents and verify their correctness by using UPPAAL.

We developed a simulator X-SH which can be used

by both home users and providers in order to estimate

and manage the daily consumptions. This tool was

tested at Cynapsys and some experimental results are

exposed in this paper. In our future work, Cynapsys

plans the real commercialisation of this product. We

are planning in this case to ﬁnish the complete de-

ployment of this simulator as well as Agents on mi-

crocontrollers following STM32F4 technology. The

real commercialisation and distribution is planned for

the end of 2014.

REFERENCES

Abras, S., Ploix, S., Pesty, S., and Jacomino, M. (2008). A

multi-agent home automation system for power man-

agement. In Springer Berlin Heidelberg, pages 59–68.

AlShu’eili, H., Gupta, G. S., and Mukhopadhyay, S. (2011).

Voice recognition based wireless home automation

system. In 4th International Conference on Mecha-

tronics (ICOM), pages 1–6.

Alur, R. and Dill, D. L. (1994). A theory of timed automata.

In Theoretical computer science, pages 183–235.

Angelov, C., Sierszecki, K., and Marian, N. (2005). De-

sign models for reusable and reconﬁgurable state ma-

chines. In L.T. Yang et al. Eds. Proc. Of Embedded

Ubiquitous Comput, pages 152–163.

A.Vichare and Verma, S. (2012). Embedded web server for

home appliances. In National Conference on Emerg-

ing Trends in Engineering and Technology, VNCET.

Bengtsson, J., Larsen, K., Larsson, F., Pettersson, P., and

Yi, W. (1996). Uppaal: a tool suite for automatic ver-

iﬁcation of real-time systems. In Hybrid Systems III,

pages 232–243.

Debono, C. J. and Abela, K. (2012). Implementation of a

home automation system through a central fpga con-

troller. In Electrotechnical Conference (MELECON),

16th IEEE Mediterranean, pages 641–644.

SMARTGREENS2014-3rdInternationalConferenceonSmartGridsandGreenITSystems

384

Gill, K., Yang, S., Yao, F., and Lu, X. (2009). A zigbee-

based home automation system. In IEEE Transactions

on Consumer Electronics, pages 422–430.

Ha, Y. G. (2009). Dynamic integration of zigbee home net-

works into home gateways using osgi service registry.

In IEEE Transactions on Consumer Electronics, pages

470–476.

Hu, L., Lee, K.-M., Zou, J., Fu, X., and Yang, H.-Y.

(2011). Adaptive measurement for automated ﬁeld re-

construction and calibration of magnetic systems. In

IEEE Transaction, pages 327–337.

Khalgui, M., Mosbahi, O., Li, Z. W., , and Hanisch, H. M.

(2011). Reconﬁgurable multi-agent embedded control

systems: From modelling to implementation. In IEEE

Transaction, pages 538–551.

Megherbi, D. B. and Madera, M. (2010). A hybrid p2p and

master-slave architecture for intelligent multi-agent

reinforcement learning in a distributed computing en-

vironment: A case study. In IEEE Transaction.

Nunes, R. J. C. (2010). Domobus-a new approach to

home automation. In 4th International Conference on

Mechatronics (ICOM).

Nwana, H. S., Lee, L., and Jennings, N. R. (1996). Co-

ordination in software agent systems. In BT Technol

Palshikar, G. K. (2004). An introduction to model checking.

In Tata Research Development and Design Centre.

Rooker, M. N., Sunder, C., Strasser, T., Zoitl, A., Hum-

mer, O., and Ebenhofer, G. (2007). Zero downtime re-

conﬁguration of distributed automation systems: The

cedac. In Third International Conference on Indus-

trial Applications of Holonic and Multi-Agent Sys-

tems.

R.Piyare and M.Tazil (2011). Bluetooth based home au-

tomation system using cell phone. In IEEE 15th In-

ternational Symposium on Consumer Electronics.

Seokcheon, L. (2010). Fairness, stability and optimality of

adaptive multiagen systems. In Interaction through

resource sharing, IEEE Trans, pages 427–439.

Shunyang, Z., Du, X., Yongping, J., and Riming, W. (2007).

Realization of home remote control network based on

zigbee, electronic measurement and instruments. In

ICEMI 07, 8th International Conference, pages 344–

348.

Song, G., Wei, Z., Zhang, W., and Song, A. (2007). Design

of a networked monitoring system for home automa-

tion. In Consumer Electronics, IEEE Transactions,

pages 933–937.

STMicroelectronics (2013). Datasheet. In STM32F4

Datasheet.

Torbensen, R. (2008). Ohas open home automation system,

consumer electronics. In ISCE IEEE International

Symposium on Consumer Electronics, pages 1–4.

Wang, X., Khalgui, M., Li, Z., and Mosbahi, O. (2010). Au-

tomatic low-power reconﬁgurations of real-time em-

bedded control systems. In Technical Reprot Systems

Control and Automation Group School of Electro-

Mechancial Engineering Xidian University.

NewSoftwareSolutionsforLow-powerManagementofGreenSmartHomes

385