Adaptive STM32F4 Microcontrollers

Case of Flexible Smart Meters

∗

Aymen Jaouadi

1,2

, Olfa Mosbahi

, Mohamed Khalgui

and Ahmed Toujani

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 Meter, Smart Grid, Software Agent, Modeling and

Veriﬁcation, Simulation.

Abstract:

The research paper deals in theoretical level with ﬂexible and adaptive microcontrollers following the well-

known industrial technology STM32F4. It is applied in the practical level to a Smart Meter SM which is

developed at Cynapsys for future generations of Smart Grids. A reconﬁguration scenario in theory is assumed

to be any ﬂexible operation allowing the addition-removal-update of OS tasks in order to adapt the micro-

controller to its environment according to user requirements. It is assumed in practice to be any addition-

removal-update of new services to-from SM such as the energy consumption, the remote information reading

and power shutdown, the stabilization of the delivered power, the management of new power provider offers,

the sale of energy and ﬁnally the peak consumption management. We propose an agent-based architecture for

a STM32F4 device where a hierarchical software agent is deﬁned to control the environment evolution before

applying local reconﬁgurations for a required ﬂexibility of the microcontroller. We model the agent by using

nested timed automates, and design the whole architecture to manage all possible reconﬁguration forms. The

agent-based architecture is totally implemented and applied to SM, and a simulator X-SM is developed for the

evaluation of this paper’s contribution.

1 INTRODUCTION

Microcontrollers, STM32 F4 are special-purpose

computer systems designed to perform one or few

dedicated functions often with real-time computing

constraints in order to control physical processes in

the real world. The requirements in the development

of microcontrollers are increasingly growing in term

of ﬂexibility and agility (Vyatkin et al., 2005) (Pratl

et al., 2007). The most promising directions to ad-

dress these issues is the reconﬁguration of one of

the most used microcontrollers in the world which

is the STM32 F4 to be autonomous as a new chal-

lenge, STM32 F4 represents 45% of the microcon-

troller market, and provides users a number of an-

∗

This research work is a collaboration between LISI

Laboratory (INSAT Institute) at University of Carthage and

the German-French Company Cynapsys in Tunisia. We

thank Ing. Haythem Tebourbi Technical Director and Ing.

Souhail Kchaou Director of Research and Development for

long fruitful discussions and stable ﬁnancial and technical

supports.

nexes integrated component, it allows the addition

of devices, such as the management of LCD dis-

plays. This functionality refers to the process of

modifying the software as well as hardware struc-

ture and behavior of the system during its execution.

Being reconﬁgurable is important for reacting fast

to sudden and unpredictable requirement changes or

perturbations with minimum cost and risk. Several

interesting academic and industrial research works

have been made in recent years to develop recon-

ﬁgurable control systems (Gehin and Staroswiecki,

2008).We distinguish in these works two reconﬁgu-

ration policies: static and dynamic reconﬁgurations

such that static reconﬁgurations are applied off-line

to apply changes before the system could starts (An-

gelov et al., 2005), whereas dynamic reconﬁgurations

are applied dynamically at run-time. Two cases ex-

ist in the last policy: manual reconﬁgurations ap-

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

reconﬁgurations applied by Agents (Al-Saﬁ and Vy-

atkin, 2007)(Theiss et al., 2009). We are interested

in software reconﬁgurations, and assume that all the

364

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

Adaptive STM32F4 Microcontrollers - Case of Flexible Smart Meters.

DOI: 10.5220/0004975903640374

In Proceedings of the 3rd International Conference on Smart Grids and Green IT Systems (IEEHSC-2014), pages 364-374

ISBN: 978-989-758-025-3

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

reconﬁgurations are applied automatically by an in-

ternal module of the microcontroller that should be

kept autonomous as much as possible. In the case of

a manual reconﬁguration, we assume that the recon-

ﬁguration agent interacts with user to execute there

requests. This controller is assumed to be a superset

of OS software tasks such that only a particular subset

is loaded at run-time to implement it under user con-

straints. A reconﬁguration will be any automatic op-

eration allowing the addition, removal and update of

tasks at run-time. We propose a new module named

Embedded Control Module ECM to handle software

reconﬁgurations of the microcontroller. This mod-

ule is a superset of sub-modules that manage all pos-

sible reconﬁguration scenarios such as the addition,

removal or update of tasks at run-time (Zhang and

Wang, 2010) (Pfeffer and Ungerer, 2004) (Chauhan

et al., 2009) (Otero et al., 2006). We propose a UML-

based model for this ECM which plays a key func-

tion inside the autonomous microcontroller. We apply

this contribution by assuming the case of electric me-

ters following the STM32 technology of microcon-

trollers. We consider in particular the smart meter

developed at the German-French Company Cynapsys

located in Tunisia. This meter supports the classic

task dealing with the computation of the energy con-

sumption. Nevertheless, the company plans to add

new functionalities that allow these meters to be au-

tonomous and smart. The remote shutdown of the

electric power when the bill is not paid, the stabiliza-

tion of energy, the remote reading of data or display of

news, the management of peaks, the management of

renewable energy or the remote negotiation between

the provider and the clients, will be new original ser-

vices to be added to the microcontrollerof the planned

Smart Meter in order to enhance the performance of

the devices and optimize the energy management. All

these original services to be considered as OS tasks

will be handled by the ECM that adapts the device ac-

cording to user requirements by adding or removing

or updating tasks at run-time. The planned smart me-

ter will be a new challenge for Cynapsys in Tunisia to

guarantee services with high qualities and with low-

costs. We specify these services as well as the ECM

according to the model checker by using the tool UP-

PAAL which is used to simulate the new meter (Alur

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

also the implementation of the STM32 smart meter

and show the experimentations that we did at Cynap-

sys to evaluate the whole contribution. The organisa-

tion of this paper is as follows. The next section an-

alyzes the Background and a detailed description of

the Reconﬁgurable Embedded Systems, the STM32

Microcontrollers and Current Smart Meters. Section

3 is dedicated to the description of the Smart Meter at

Cynapsys Company and describe new challenges and

solutions. Section 4, proposes reconﬁgurable STM32

microcontrollers where we formalize, model and de-

sign their behaviors followed by a detailed model-

ing of our system modules by using UML Diagrams.

Section 5 proposes the new original Smart Meter at

Cynapsys where we detail their veriﬁcation, design

and implementation. Finally Section 6 concludes this

work. Our approach is original by implementing new

services in the Smart Meter to manage production and

the optimisation of the consumption at Real-Time.

2 BACKGROUND

In this section, we present an overview on recon-

ﬁgurable embedded systems, the STM32 microcon-

trollers family and the current used smart meters.

2.1 Reconﬁgurable Embedded Systems

An embedded system is reconﬁgurable if it changes

its software or hardware behavior at run-time accord-

ing to user requirements. The software reconﬁgura-

tion is any operation allowing the addition, removal

or update of software tasks that implement the sys-

tem to encode corresponding functions. The hard-

ware reconﬁguration is assumed to be any operation

allowing the addition, removal or update of hardware

components according to user requirements. An ad-

dition or removal can be of memory, of data-event

inputs-outputs, or of a new network for communi-

cation. The update of hardware components can be

the modiﬁcation of the processor speed. The con-

stant growth of the complexity afferent and neces-

sary to the management of embedded software sys-

tems makes reconﬁguration autonomy increasingly

important. The challenges include both the model de-

sign level and the environment level of runtime sup-

port. According to(Boukhannoufa, 2012), real-time

systems can be large, distributed, and have a dynamic

environment. This requires the introduction of var-

ious modes of operation and reliability techniques

to ensure its operation and maintainability. More-

over, these dynamic changes of architecture and be-

havior have a negative impact on the temporal char-

acteristics of systems that require a special study on

the ability of adaptive behaviors to ensure the hard

real-time constraints imposed to the systems. These

adaptive behaviors amplify the complexity of devel-

oping real-time systems. According to (Wang et al.,

2010), the new generations of embedded control sys-

tems are addressing new criteria such as ﬂexibility

AdaptiveSTM32F4Microcontrollers-CaseofFlexibleSmartMeters

365

and agility. To enhance their operations, the embed-

ded control systems should be changed when ran-

dom disturbances happen or when improvements of

the performance should be applied and propose a dy-

namic low power reconﬁgurations of real-time em-

bedded control systems that should respect hard real-

time constraints and perform modiﬁcations of the pe-

riods and deadlines, modiﬁcation of the worst case

execution times (WCETs), and ﬁnally the removal of

some tasks to minimize the energy consumption in or-

der to adjust the behavior of the processor. According

to (Kramer and Magee, 1985), the authors propose

a reconﬁguration model based on the dynamic incre-

mental modiﬁcation and extensions. In this work,

the required properties are determined by languages

and their execution environment. In (Thramboulidis,

2004b) (Thramboulidis, 2004a), the authors propose a

new vision for the reconﬁguration of distributed con-

trol applications. The main objective is to bridge their

works on software engineering (eg. UML) to IEC

61499 (J.H.Christensen et al., 2005). In (Zhang et al.,

2013) the authors present a reconﬁguration architec-

ture for real-time distributed control systems. Specif-

ically, the low-level control of the physical compo-

nents which is shown with the handling of real-time

requirements. The main contribution of this work is to

deﬁne a reconﬁguration request that interacts with the

application under reconﬁgurations through three in-

terfaces. More speciﬁcally, the modiﬁcation interface

provides its role based on a set of key reconﬁgura-

tion services identiﬁed in a previous work (Zoitl et al.,

2010). Our contribution is to provide an original ap-

proach for the automatic and manual agent-based re-

conﬁguration model for STM32 F4 microcontrollers

implementing original services for low power man-

agement.

2.2 STM32 Microcontrollers

A microcontroller is an integrated circuit that brings

together the essential elements of a computer: CPU,

memory (ROM for program RAM for data), periph-

eral units and input-output interfaces. Microcon-

trollers are characterized by a higher degree of inte-

gration, lower power consumption, lower operating

speed, and reduced cost compared with versatile mi-

croprocessors used in personal computers. Microcon-

trollers are widely used in embedded systems such

as auto engine systems, remote controls, home appli-

ances and mobile phones. The STM32 technology of

32bit Flash microcontrollers based on the ARM Cor-

tex processor is proposed to provide a 32bit product

range that combines high performance, real-time ca-

pabilities, digital signal processing, and low-power,

low-voltage operation, while maintaining full integra-

tion and ease of development. The different STM32

microcontrollers, based on an industrial standardiza-

tion with a large choice of tools and software, make

this family of STM32 the ideal technology for small

projects and for entire platform decisions. To apply

our reconﬁguration approach, we opted for the choice

of a STM32 card type STM32 F4 Discovery which

is based on STM32F407 Cortex M4 controller with

1MO Flash and 192KO RAM. The STM32F407 fam-

ily is based on the high-performance ARM Cortex-

M4 32-bit RISC core operating at a frequency of up

to 168 MHz. The Cortex-M4 core features a ﬂoat-

ing point unit (FPU) single precision which supports

all ARM single precision data-processing instructions

and data types. It also implements a full set of in-

structions and a memory protection unit (MPU) which

enhances application security. This card uses power

from the USB bus or from an external 5V power

supply, a 3-axis accelerometer LIS302DL, two but-

tons and a micro USB connector (STMicroelectron-

ics, 2013). The use of STM32 F4 cards is the ﬁrst ap-

proach in the ﬁeld of reconﬁgurable microcontrollers,

in terms of originality, our scientiﬁc paper is charac-

terized by the application of different types of archi-

tectural, scheduling and data reconﬁguration in order

to add new services at runtime to the operation of the

Smart Meter developed at Cynapsys.

Figure 1: STM32 F4 Discovery Card.

2.3 Overview on Current Smart Meters

According to (S.Depuru et al., 2011), a smart meter

is an advanced energy meter that measures the con-

sumption of electrical energy, provides additional in-

formation compared to a conventional energy meter.

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

366

A smart meter is a counter that automatically registers

and communicates data on electricity consumption, it

is divided into two main categories: traditional meters

(the old electromechanical meters) measure only the

total electricity consumption and Smart meters mea-

sure the amount of electricity consumed and the time

of consumption, they automatically transmit data to

providers by using a wireless communication technol-

ogy (Ahmad, 2011) (Hoenkamp and Huitema, 2012).

The price of electricity with these data may vary dur-

ing the day, giving houses a new way to manage the

costs of their consumptions. You could, for exam-

ple, choose to reduce your consumption during the

most expensive time (peak or normal) and partially

use electricity during off-peak hours, during which it

will be cheaper. Investigations conducted in Tunisia

have for the moment appear incomprehensible bill in-

creases in number. These Meters are intelligent and

used to regulate the energy delivery in peak of con-

sumption periods, for example the French Electricity

Distribution Network ERDF has with the Smart re-

mote Meter Linky, the ability to change their rates

from one minute to another, according to the con-

sumption of international prices electricity (ERDF,

2009). In Cynapsys, we propose a new strategy to add

the original proposed services to the old developed

Smart Meter which was a remote control example of a

DC motor. This last represents one of the end devices

that can be used at home or a building. The com-

munication with the Energy Counter is established by

means of wireless communication protocol ZigBee.

The Energy counter module allows providing infor-

mation about the average of the energy consumption

of all the elements connected to it. In addition it per-

mits to get the real-time consumption of the home or

building via ZigBee protocol. Today the deployment

of Smart Meters should allow a better understanding

of the positions of power consumption and thus save

money. Monitoring real-time also opens the door to

energy billing in real time, but also to the differenti-

ated based billing electricity demand (peak load). Into

this research we will enhance the behavior of the elec-

tric meter by the services described above in order to

make it more intelligent. Intelligence brought many

advantages for the customer and energy companies.

Starting with a bill which can be calculated on the

basis of real consumption, interventions carried out

remotely (without constraint of appointment) includ-

ing stabilization of the varying voltage and regulating

consumption by micro-cuts in periods of high con-

sumption known to peak loads. Consume less power

and better is the goal of our study. These optimiza-

tions will be applied to the electric meter developed

by Cynapsys. The intelligence provided to the me-

ter is by adding new services that provide an immedi-

ate beneﬁt to the counter in real-time. Smart meters

equipped consumers should not pay an estimate bill

but they consider real power consumptions. This new

generation of meters can display power consumption

and remote meter reading. Thus the consumer can

also save money by better monitoring its power con-

sumption. The use of smart meter certainly makes

life easier for users, thanks to its new services such

as the automatic control of electrical devices at home

and the regulation of supply and demand through re-

mote micro-cuts in order to not exceed a ﬁxed con-

sumption threshold. Our paper consists of two novel

approaches, the ﬁrst is a theoretical approach to en-

hance the electric meters with new original services,

and in the second we propose a practical approach by

proposing the X-SM Simulation tool.

3 ADAPTIVE SMART METERS

OF CYNAPSYS COMPANY:

NEW CHALLENGES AND

SOLUTIONS

At the German-French Company Cynapsys, the cur-

rent research paper is applied to a well-developed

STM32-based electric meter for the remote control

of home end devices. The meter encodes a main

counter module ECM to provide the real-time energy

consumption of any connected device to it by using

a Zigbee-based protocol. In addition to this classic

function, we propose to enrich the meter with original

services such as the remote shutdown of power, the

remote reading of the energy consumption from the

meter to the provider, the smart stabilization of power

by using renewable energy that we assume available

at home, the display of the energy consumption for

users at home, the smart management of promotions

to be offered from the provider by relatively activat-

ing or deactivating home end devices when the price

of power (per hour) is low or high, the sale of energy

to the provider when the available renewable energy

is enough, the smart peak management by deactivat-

ing home end devices during peak times, and ﬁnally

the management of news to be sent from the provider

about the network status in the next weeks or months.

These optimizations will be applied to the electric me-

ter developed by Cynapsys. Note that all these ser-

vices will not be loaded together on memory in order

to not overload the memory by the applications and

the tasks that will not be executed. Thus, the system

preserves all its resources to treat most substantial ser-

vices. Except the main service dealing with counting,

AdaptiveSTM32F4Microcontrollers-CaseofFlexibleSmartMeters

367

we assume that each another will be loaded on mem-

ory when needed at run-time. Reconﬁguration in our

approach keeps the system available, increase operat-

ing efﬁciency and especially without remarkable per-

formance degradation. We propose a model of dy-

namic automatic reconﬁguration, or possibly manual

to be executed through a decision module that handles

these cases in real-time. The current paper deals with

the theoretical level of the STM32 Microcontroller to

be assumed ﬂexible and autonomous. Each service is

assumed to be implemented by an OS task. The sys-

tem is then implemented by a superset of tasks such

that only one subset implements it at a particular time

after a well-deﬁned reconﬁguration scenario. The pa-

per deals in the application level with this Smart Me-

ter of Cynapsys to be enriched with these new ser-

vices. Each addition-removalof a service on the smart

meter corresponds to a reconﬁguration scenario. The

services and the tasks in our approach are executed

according to the priority to be assigned and based on

the importance of the operations as the remote power

shutdown that must be executed upon his arrival. We

denote in the following by:

• T

: the main software task that measures the en-

ergy consumption by reading directly and not by

an estimation of consumption. This task has the

highest priority,

• T

: the software task allowing the remote shut-

down. The meter we conceive is remotely pro-

grammable and equipped with a remote switching

device called ”AMM” (Advanced Meter Manage-

ment). This service is crucial in the case of non-

payment of bills,

• T

: the software task allowing the remote reading

from the smart meter to the power provider and

remote users,

• T

: the software task allowing the stabilization of

the delivered power from the provider by using

if possible (if the input load is decreased at run-

time) the renewable energy that we assume avail-

able at home,

• T

: the software task allowing the display of in-

formation on the meter which is equipped with a

digital displayer,

• T

: the software task allowing the management of

promotions to be offered from the provider, this

task informs users by SMS and emails about these

offers,

• T

: the software task allowing the sale of energy

to the provider when the available local energy

is enough, this task informs users by SMS and

emails,

• T

: the software task allowing the management of

peaks by using useful information to be sent by

SMS and emails from the provider,

• T

: the software task allowing the management of

all news to be sent from the provider about the

network status. This task informs users by SMS

and emails.

The proposed tasks are considered as a new gener-

ation of energy production and consumption. They

represent a solid medium for effective management

of the power across the electric Smart Meter which is

the entrance of the smart grid.

4 THEORETICAL

CONTRIBUTION:

RECONFIGURATIONS OF

STM32F4

MICROCONTROLLERS

We aim in this section to dynamically reconﬁgure a

STM32 microcontroller which is assumed to be im-

plemented by a set of independent OS tasks. The

goal is to adapt its behavior at run-time to its envi-

ronment according to well-deﬁned user requirements.

A reconﬁguration scenario is assumed to be any dy-

namic operation allowing the addition, removalor up-

date of tasks to-from the microcontroller. We pro-

pose an Embedded Reconﬁguration Module to be de-

noted by ERM that controls the evolution of the mi-

crocontroller’s behavior and considers also user re-

quirements to apply run-time reconﬁgurations. The

ERM is assumed to be encoded in three hierarchi-

cal software levels: (a) Architecture Level (to be de-

noted by AL), (b) Scheduling Level (to be denoted

by SL), and (c) Data Level (to be denoted by DL).

We deﬁne in AL, all the possible software architec-

tures that can implement the STM32 microcontroller

at run-time. An architecture in AL is a set of OS tasks

that perform control activities. A reconﬁguration sce-

nario can change the software architecture of the mi-

crocontroller by adding or also removing OS tasks.

For each architecture in AL, we need to deﬁne an ex-

ecution model of the corresponding tasks. A schedul-

ing is then deﬁned in SL to affect a priority to each

task. For each architecture and for each scheduling

of the corresponding tasks, we deﬁne also in DL all

the possible corresponding values of data to be han-

dled at run-time. Thanks to this hierarchical structure,

the ERM can handle all possible reconﬁguration sce-

narios of a STM32 microcontroller, our approach is

original because no previous work did the same hier-

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

368

archical architecture on STM32F4 to handle reconﬁg-

uration scenarios.

4.1 Formalization of Reconﬁgurable

STM32 Microcontrollers

Let Sys be the STM32 microcontroller that can be re-

conﬁgured at run-time to adapt its behavior to its en-

vironment. We denote by Γ

Sys

the big set of all the

possible tasks involved in the different implementa-

tions of the system Sys, which is implemented at any

particular time t by a subset ξ

Sys

which represents the

set of tasks involved in a particular implementation

Sys

⊆ Γ

Sys

. We model the architectural level AL of

ERM by a ﬁnite state machine S

such that each state

of S

corresponds to a particular implementation at

architectural level.

= (Γ

Sys

,O,δ), such that,

• O is a set of n states in S

(O = {S

/i ∈ 1..n}),

• δ is a state-transition function Γ

Sys

XO → Γ

Sys

A reconﬁguration scenario R

i, j

is a transition from

a state S

corresponding to particular subset of tasks

Sys

to a state S

corresponding to particular subset

of tasks ξ

Sys

. In this scenario, we assume run-time op-

erations allowing addition-removal of tasks to adapt

the system Sys to its environment. For each state S

we deﬁne in the second hierarchical level(Scheduling

Level SL) a particular state machine to be denoted by

. Each state in S

deﬁnes a particular scheduling

of the subset of tasks ξSys. This scheduling affects a

priority to each task in order to get a deterministic ex-

ecution model of the microcontroller Sys. We denote

by Ψ(ξ

Sys

) the set of all possible execution models of

tasks of ξ

Sys

at the Scheduling Level.

= (Ψ(ξ

Sys

),P,β), such that,

• P is a set of m scheduling states in S

(P =

/i ∈ 1..m})

• β is a state-transition function Ψ(ξ

Sys

)XP →

Ψ(ξ

Sys

)XP

A reconﬁguration scenario R

i, j

at Architectural

Level AL, is a transition from a state S

to another

state S

of S

. The reconﬁguration of the microcon-

troller Sys in the third hierarchical level DL can be the

update of data. We deﬁne for each state S

of S

and also for each state S

of S

a new state machine

where each state corresponds to new values to be

affected to data of tasks belonging to ξ

Sys

under the

scheduling S

. Let ϒ(ξ

Sys

) be the set of all possible

values of data for the tasks of ξ

Sys

under the schedul-

ing S

= (ϒ(ξ

Sys

),Q,β), such that,

• Q is a set of l data states in S

(Q = {S

/i ∈

1..l})

• β is a state-transition function ϒ(ξ

Sys

)XQ →

ϒ(ξ

Sys

)XQ

A reconﬁguration scenario R

k,h

is a transition to

change the values of data from a state S

to another

state S

of S

. We denote ﬁnally by Con fig

i, j,k

conﬁguration of the STM32 microcontroller Sys to be

implemented by the subset of tasks ξ

Sys

under a well-

deﬁned scheduling corresponding to a state S

in S

with particular values of data deﬁned in a state S

in S

. A reconﬁguration scenario to be denoted by

Reconfig

u,v,w

i, j,k

of Sys is deﬁned then as any run-time

operation allowing the adaptation of the microcon-

troller from a conﬁguration Config

i, j,k

to a new one

Config

u,v,w

. The originality of our approach is distin-

guished by a speciﬁc formalization of the scenarios

as possible reconﬁgurations and the exclusivity of the

application of our original ideas to the Electric Smart

Meter.

4.2 UML-based Design of

Reconﬁgurable STM32 Micro

Controllers

To implement the different run-time reconﬁguration

scenarios of the STM32 microcontroller Sys, we pro-

pose in Figure 2 a UML class diagram that designs

both the Embedded Control Module ECM and also

the different OS tasks belonging to Γ

Sys

. Our class

diagram consists of a ECM class that plays a very

important role in our reconﬁguration approach. This

class manages all the interactions in the system, it

is related to the classes AL, SL, DL which respec-

tively represent the architectural reconﬁgurationlevel,

scheduling reconﬁguration level and data reconﬁgura-

tion level. We have also four other classes: Provider

Interface, User Interface, Equipment Interface and

Task. The Task class represents all the tasks included

in the different implementations of our system. The

user interface class represents all the customers of our

system. The Equipment Interface class, includes all

devices connectedto the counter. Finally, the Provider

Interface class manages the relationship with the elec-

tricity supplier.

AdaptiveSTM32F4Microcontrollers-CaseofFlexibleSmartMeters

369

Figure 2: UML Class Diagram of an Adaptive STM32 Mi-

crocontroller.

5 PRACTICAL CONTRIBUTION:

CASE OF NEW SMART

METERS AT CYNAPSYS

In this section we present the practical contribution of

our approach.

5.1 Timed Automata Models

We describe in Figure 3 the different state machines

encoding the ECM of the STM32-based Smart Meter.

The state machine encoding the Architectural Level is

composed of four states:

• Counting the energy consumption

• Regulation of the energy

• Promotion negotiation

• Remote shutdown of the electricity

The ﬁrst state Counting corresponds to the sys-

tem’s implementation ξ

Sys

= {T

} when the

main task T

is active as well as the tasks T

that

allow respectively the remote reading, the display and

the management of news. The second state Regula-

tion corresponds to the system’s possible implemen-

tation ξ

Sys

= {T

} when the main task T

also active as well as the tasks T

that allow re-

spectively the stabilization, the display and the man-

agement of news. The third state Negotiation corre-

sponds to the system’s possible implementation ξ

Sys

} when the main task T

is active

as well as the tasks T

that allow respec-

tively the display, the management of promotions, the

sale of energy, the management of peaks, and ﬁnally

the management of news. We describe also in the

scheduling levelthe different possible execution mod-

els of tasks encoding each state in the architectural

level. The scheduling Sched1 of the state Counting

executes the main task of counting T

each time it ex-

ecutes T

, T

and T

. The precision of counting is

high in this case. The scheduling Sched2 of the same

state Counting reduces the precision by executing T

two times in a trace of execution. The precision of

counting is low in Sched3 that executes T

only one

time. We describe in Data Level the periodicity of

each trace of execution in the second level. In this

case, Sched1, Sched2 and Sched3 are respectively ex-

ecuted periodically each 100ms, 300ms and 500ms.

Figure 3: State Machines of the Embedded Control Module

that handle the reconﬁgurations of the STM32-based Smart

Meter.

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

370

The design of reactive systems must comply with

logical correctness: ”the system does what it is sup-

posed to do” and timeliness: ”the system has to sat-

isfy a set of temporal constraints”. The smart meter

should respect the notion of time such as if a power

interruption is requested it should execute this task

immediately, then it can reactivate the electricity due

to the existence of functional constraints such the pay-

ment of bills or the passage of a period of high con-

sumption. We model and simulate the different state

machines of ECM as well as the different assumed

tasks by using the well-known model checker UP-

PAAL in order to check the behavior of the adaptive

smart meter after each reconﬁguration scenario. Ac-

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

most successful approach that’s emerged for verify-

ing requirements. We describe in Figure 4 the model

of the second task T

allowing the remote shutdown

of the smart meter according to the formalism Timed

Automata. The meter we conceive is remotely pro-

grammable and equipped with a remote switching de-

vice called ”AMM” (Advanced Meter Management).

This service is crucial in the case of non-payment of

bills, the deployment of a team to ensure power cuts

today becomes useless.

Idle

treshold=newTreshold(), y:=0

NotResolved

RemoteShutdown

!Bill

y<=15

Resolved

energyMetering()>=treshold y>15

y>15

[paidbill]

Figure 4: Model of T

The voltage variation is deﬁned in relation to other

disturbances that can affect the electric network. It

is generally deﬁned as a decrease followed by an in-

crease (sometimes the opposite, but less frequently)

of the electric intensity. We describe in Figure 5 the

model of the task T

allowing the stabilization of the

consumed energy. The operation of stabilization is

based on the use of the stored energy derived from

green renewablesources. In the case of voltage ﬂuctu-

ation the battery is checked in the state BatteryCheck

and the compensation process begins.

activeSolution[ID]

Brownout

StopCompensation

handleResponse()

isFluctuation[ID]?

BatteryCheck

BatteryCompensating

VoltageFluctuation

Idle

startCompensation()

haveAvailableLoad()

EquipementStop

VoltageFluctuation

VoltageRise[over]

VoltageRegulated

Figure 5: Model of T

5.2 Implementation of the X-SM

Simulation Tool

The simulation is a mandatory step for the ﬁnal de-

ployment of any system to be represented by a model.

In our research work, a simulator is developed for

the simulation of all these services that encode the

electric meter of our company Cynapsys. We devel-

oped an environment X-SM for the simulation of a

STM32F4 based smart meter. Our tool is essentially

a program that allows the counting (a)the total power

consumption of all devices that are connected to the

electric meter, (b) the execution of any remote shut-

down of the electrical energy in the case of exceeding

the threshold of consumption, (c) the management of

promotional offers proposed by producers,(d) and ﬁ-

nally the stabilization of the energy variation. The

main interface of our tool presents the different ser-

vices offered by the Smart Meter and the various de-

vices connected to it. To test the microcontroller, we

assume three types of devices: a lamp, a refrigerator

and a boiler with their characteristics and parameters

such as the device name, the power, the device state

(on or off), the priority, the shutdown condition, the

output number (identiﬁer) and the period of use.

The power consumption counting service, oper-

Figure 6: Tool Main Interface.

AdaptiveSTM32F4Microcontrollers-CaseofFlexibleSmartMeters

371

ates in real-time and provides the equivalent amount

of the consumed quantity. It stills compared to the

ﬁxed consumption threshold as shown in the ﬁgure

below.

Figure 7: Energy Counting Service Interface.

The remote shutdown service interface informs

the user by SMS and email, that the consumption

threshold is exceeded. It performs the shutdown op-

eration and changes the state of the device from On to

Off. The user is informed in parallel about the elec-

tricity shutdown so that he can pay his unpaid bills, or

rationalize his energy consumption.

Figure 8: Remote Shutdown Service Interface.

The voltage stabilization is a capital service, it

helps to maintain the electrical equipment in the

house against voltage variations that can damage

them. The interface of the tool detects a variation and

identiﬁes its type, whether it is an increase or decrease

in voltage. It informs the user by SMS and email that

a shutdown must be performed. After the Shutdown

the system pass to the compensation solution by using

the amount of energy stored in the batteries produced

from supposed renewable sources at home.

Below the management peak consumption service

interface proposed by our environment X-SM, a peak

energy consumption is generated by a strong demand

for electricity and an imbalance in supply and de-

mand. The smart meter allows smooth management

of these peak periods through an anticipation mecha-

nism.

Figure 9: Voltage Stabilization Service Interface.

Figure 10: Peak Consumption Management Service Inter-

face.

After crossing the period of peak consumption,

the user regains his normal operating mode. The

stopped devices will resume there normal executions.

The system of anticipation / reaction offers an innova-

tive solution that provides a huge gains and preserve

the equipment in houses, which can be damaged by

sudden and unexpected cuts. Below the end of the

peak consumption interface.

5.3 Experimentation

The ﬁgure 12 shows a comparison between the results

obtained before and after the contribution of our pro-

posal. We assume that the daily consumption of elec-

tricity in regular time is estimated at 4 KWH. Dur-

ing peak consumption this value should not exceed

2 KWH per day. The reconﬁguration applied to our

electric meter, gives rise to direct cuts that should not

exceed the ﬁxed amount of electricity.

After connecting the STM32 F4 card to our X-

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

372

Figure 11: End of Peak Consumption Service Interface.

0 5 10 15 20 25

0.5

1.5

2.5

3.5

hours

KWH

Before contribution

After contribution

Figure 12: Peak Consumption Management Service.

SM simulator, the data acquisition is performed via a

communication and information collecting protocol.

Our simulation environment shows the reaction of the

electric smart meter face to the detection of voltage

variation. The reconﬁguration process enables the

service of voltage stabilization via the compensation

through renewable production sources and electrical

equipment shutdown, so they are not damaged.

6 CONCLUSIONS

This paper deals with an agent-based dynamic re-

conﬁguration of a microcontroller solution. Our ap-

proach is started by a complete study of the actual

electric meter in the ﬁeld of energy development and

applied to the Cynapsys developed smart one. The

reconﬁguration agent allows the embedded systems

to change its software or hardware behavior at run-

time according to user requirements or environmen-

tal changes. We propose in this work an embedded

Figure 13: Energy Stabilisation Histogram.

reconﬁguration of the microcontrollers following the

STM32 F4 technology the well-used in the world.

To manage the reconﬁguration process we propose

an agent-based architecture, composed of the archi-

tectural level, Scheduling level, and data level. We

present a detailed formalization, veriﬁcation of nested

state machine models and a software design version

which is based on UML Class Diagram before the de-

velopment of complete simulation tool applied to the

Smart Meter developed at the German-French com-

pany Cynapsys. The proposed approach is original

and is distinguished from the related works in this

ﬁeld, the product will now proceed to the production

and marketing by the company Cynapsys. we will fo-

cus more in the future on the communication between

electricity providers and smart meters, we will give

more importance to the electricity metering based on

the real consumption of the consumer, deploy this

prototype in a smart city and ﬁnally the massive in-

dustrialization of this innovative product.

REFERENCES

Ahmad, S. (2011). Smart metering and home automation

solutions for the next decade. In International Confer-

ence on Emerging Trends in Networks and Computer

Communications (ETNCC), pages 200–204.

Al-Saﬁ, Y. and Vyatkin, V. (2007). An ontology-based

reconﬁguration agent for intelligent mechatronic sys-

tems. In Third International Conference on Industrial

Applications of Holonic and Multi-Agent Systems.

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 International Federation for Information

Processing, pages 04–17.

AdaptiveSTM32F4Microcontrollers-CaseofFlexibleSmartMeters

373

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.

Boukhannoufa, M. (2012). Adaptabilit et reconﬁguration

des systmes temps-rel embarqus. In UNIVERSITE

PARIS-SUD.

Chauhan, A., Rajawat, A., and Patel, R. (2009). Reconﬁg-

uration of fpga for domain speciﬁc applications using

embedded system approach. In International Confer-

ence on Signal Processing Systems, pages 438–442.

ERDF (2009). Linky documentation. In Linky, le compteur

nouvelle gnration.

Gehin, A.-L. and Staroswiecki, M. (2008). Reconﬁguration

analysis using generic component models. In IEEE

Transactions on Systems, Machine and Cybernetics,

pages 152–163.

Hoenkamp, R. A. and Huitema, G. B. (2012). Good stan-

dards for smart meters. In 9th International Confer-

ence on the European Energy Market (EEM), pages

1–6.

J.H.Christensen, Vyatkin, V., Strasser, T., Valentini, A., and

Zoitl, A. (2005). Theiec 61499functionblockstandard:

Overviewof the secondedition. In International Elec-

trotechnical Commission.

Kramer, J. and Magee, J. (1985). Dynamic conﬁguration

for distributed systems. In IEEE Transactions on Soft-

ware Engineering, pages 424–436.

Otero, J., Wagner, F., and Carro, L. (2006). Reconﬁgura-

tion of embedded java applications, parallel and dis-

tributed processing symposium. In IPDPS, 20th In-

ternational.

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

In Tata Research Development and Design Centre.

Pfeffer, M. and Ungerer, T. (2004). Dynamic real-time re-

conﬁguration on a multithreaded java-microcontroller.

In Seventh IEEE International Symposium on Object-

Oriented Real-Time Distributed Computing, pages

86–92.

Pratl, G., Dietrich, D., Hancke, G., and Penzhorn, W.

(2007). A new model for autonomous, networked con-

trol systems. In IEEE Transactions on Industrial In-

formatics.

Rooker, M., Sunder, C., Strasser, T., Zoitl, A., Hummer, O.,

and Ebenhofer, G. (2007). Zero downtime reconﬁgu-

ration of distributed automation systems. In Third In-

ternational Conference on Industrial Applications of

Holonic and Multi-Agent Systems.

S.Depuru, Wang, L., Devabhaktuni, V., and Gudi, N.

(2011). Smart meters for power grid challenges issues,

advantages and states. In Power Systems Conference

and Exposition (PSCE), IEEE/PES, pages 1–7.

STMicroelectronics (2013). Datasheet. In STM32F4

Datasheet.

Theiss, S., Vasyutynskyy, V., and Kabitzsch, K. (2009).

Software agents in industry: A customized framework

in theory and praxis. In IEEE Transactions on Indus-

trial Informatics.

Thramboulidis, K. (2004a). Model integrated mechatron-

ics: An architecture for the model driven development

of mechatronic systems. In 2nd IEEE International

Conference on Mechatronics, pages 497–502.

Thramboulidis, K. (2004b). Using uml in control and au-

tomation: A model driven approach. In 2nd IEEE In-

ternational Conference on Industrial Informatics.

Vyatkin, V., Christensen, J., and Lastra, J. (2005). An

open, object-oriented knowledge economy for intel-

ligent distributed automation. In IEEE Transactions

on Industrial Informatics, pages 04–17.

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.

Zhang, J., Khalgui, M., Li, Z., and Mosbahi, O. (2013). R-

tnces: A novel formalism for reconﬁgurable discrete

event control systems. In IEEE Transactions On Sys-

tems, Man, And Cybernetics, Part A: Systems And Hu-

mans, pages 757–772.

Zhang, L. and Wang, Z. (2010). Design of embedded con-

trol system based on arm9 microcontroller. In Inter-

national Conference on Electrical and Control Engi-

neering, pages 3579–3582.

Zoitl, A., Lepuschitz, W., Merdan, M., and Vallee, M.

(2010). A real-time reconﬁguration infrastructure for

distributed embedded control systems. In IEEE Inter-

national Conference Emerging Technologies and Fac-

tory Automation.

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

374