diversity - DIVERSIFY

architectgroundhogInternet και Εφαρμογές Web

4 Δεκ 2013 (πριν από 3 χρόνια και 8 μήνες)

63 εμφανίσεις

DIVERSIFY

Ecology
-
inspired software diversity for distributed adaptation
in CAS


1


1
slide

about the main
idea

/ challenge


1
slide

about objectives


2
slides

about budget


1
slide

about IP


3
slides

about impact (
meta

design, adaptive
systems
, soft
diversity
)


1
slide

about WP2


1
slide

about
advances

in soft
diversity

(update
slide

24)


2

Collaborative adaptive
systems


Large
-
scale


Open


Dynamic


Eternal


Heterogeneous

environments


Face
unpredictable

situations


3

4

CASs

are a
form

of
complex

system

An essential
property
:
diversity

5

Main
idea


Diversity is an essential characteristic of
complex systems to adapt to unpredicted
changes in their environment


Ecosystems, economical systems, social systems, etc.


CASs are deployed in environments that evolve
in uncontrolled and unpredicted ways

BUT


Software diversity remains very little explored as
an insurance principle to adapt to changes

6



DIVERSIFY brings together researchers
from the domains of software
-
intensive
distributed systems and ecology in order
to translate ecological concepts and
processes into software design principles


7

Consortium

8

Ecological

board

9

M.
Hutchings


(
Univ
. of Sussex)

B.
Kunin

(
Univ
. of Leeds)

E.
Thébault

(CNRS)

C.
Melian

(EAWAG)

Objective

DIVERSIFY

aims

at

formalizing

and

experimenting

new

models

and

synthesis

mechanisms

for

software

diversity

in

collaborative

adaptive

systems,

based

on

the

ecological

concept

of

biodiversity
.

The

goal

is

to

increase

adaptive

capacities

in

the

face

of

structural

and

environmental

variations
.


10

WP structure

11

Progress in software engineering


Software
diversity


synthesis

and
spontaneous

emergence

of software
diversity


Dynamic

adaptation


leveraging

diversity

to
reach

specific

goals


Distributed

adaptation


models@runtime

for the collaboration of
heterogeneous
,
distributed

software
entities

12

Expected

impact
-

science


Genuine

ecological

inspiration for
distributed

adaptation


Continuous

evolution

and
approximate

correctness

13

i
n
t
e
l
l
i
g
e
n
t

d
e
si
g
n

p
a
ra
d
i
g
m
f
o
r
so
f
t
w
a
re

e
n
g
i
n
e
e
ri
n
g
D
I
VER
SI
F
Y
:
e
me
rg
e
n
t

d
i
ve
rsi
t
y
t
o

i
mp
ro
ve

a
d
a
p
t
a
t
i
o
n
me
t
a
-d
e
si
g
n

p
a
ra
d
i
g
m
f
o
r
so
f
t
w
a
re

e
n
g
i
n
e
e
ri
n
g
:

e
vo
l
u
t
i
o
n
a
ry
d
e
si
g
n

p
ri
n
ci
p
l
e
s
a
n
d

d
i
ve
rsi
t
y
f
o
r
re
si
l
i
e
n
ce
Expected

impact
-

society


Software
-
intensive, collaborative
systems

are
pervasive

in
our

society


DIVERSIFY
aims

at

experimenting

in smart
cities


Greater

robustness

of
other

forms

of CAS


assisted

living, emergency
systems
, etc.

14

Impact, management and
dissemination

15

Management structure

16

Budget

17

Efforts

18

IP management


Foreground

will

be

disseminated

in open
source


Details

about background
will

be

specified

in
consortium agreement


CA
is

based

on DESCA


19

I
nfrastructure

for collaboration


Social source code and document
repository


Private and public github
repositories


Shared Folder (SparkleShare
)


Private and public
wiki


Meeting White board (etherpad
)


Announcement (twitter
)


Website (drupal
)


Visio conference (INRIA visioconference bridge)

20

Work

plan

21

WP1 Ecological
modeling



Objectitves


ensure knowledge transfer from
ecologists


formalize and validate software diversity
models


formalize and validate software distributed adaptation
mechanisms


establish a tight connection with WP2 and WP3
through the collection of state
-
of
-
the
-
art models of
biodiversity and distributed
adaptation

22

WP2


Objectives:


models
of software
diversity in CASs


synthesize diversity.


lifecycle
of
diversity

23

W
P1
:

Eco
l
o
g
i
ca
l

mo
d
e
l
i
n
g
W
P2
:

So
f
t
w
a
re

d
i
ve
rsi
t
y


W
P3
:

D
i
st
ri
b
u
t
e
d

adaptation
T
2.3
Exp
o
si
n
g

d
i
ve
rsi
t
y
T

2
.
2

D
i
ve
rsi
t
y
syn
t
h
e
si
s
T
2.4
D
i
ve
rsi
t
y
l
i
f
e
cycl
e

T

3
.
3

D
i
ve
rsi
t
y
mo
n
i
t
o
ri
n
g
Po
o
l

o
f

so
f
t
w
a
re

d
i
ve
rsi
t
y
f
f
f
f
f
f
f
f
f
f
f
f
f
T

1
.
2

D
i
ve
rsi
t
y
and
sp
e
ci
a
l
i
za
t
i
o
n
WP3 Objectives and organization

24


Environment with diversity

Application 1

Application 2

Application N

Diversity
-
based Adaptation

T3.4

Diversity
-
Driven
Adaptation

Diversity Model

(model at runtime)

T3.3 Monitoring

T.3.2

WP2

WP2

Diversity

WP4

WP1


SoTA
: self
-
* Systems



Objectives


Capture
Application
Needs


Discover/Monitor
Diversity


Trigger Application
Adaptations

Work

Package 4

25

T2: Simulation and
experiment

T3: Evaluation and
report

T1: Domain analysis and
scenario

design


D4.1: Scenario
design and
system
investigation

D4.2: Smart City Simulator

D4.3: Simulator document and
analysis

D4.4: Experiment report

WP1:
Ecological Modeling

Reference for
scenarios

Provides
evaluation criteria

WP2&3:
Software diversity and
distributed adaptation

Application and
feedback

WP5
Dissemination
, collaboration and
exploitation


Main objectives


ensure

collaboration
inside

the
project


disseminate

results

outside

the
project


communicate

on the program and
particate

in the
FOCAS CA


3
tasks
:


Infrastructure
and support for
project

communication


Scientific

dissemination

and
exploitation


Collaboration

26

WP6
Management



W
ill
assure:


global
quality;


timely (and in respect with the budget) finalization of
the deliverables and reports; and


good communication, collaboration and transparency
between the partners and towards the European
Commission.


27

28

C
o
l
l
a
b
o
ra
t
i
ve

a
d
a
p
t
i
ve

syst
e
m
W
P4
:

Si
mu
l
a
t
i
o
n

o
f

d
i
ve
rsi
t
y
f
o
r
d
i
st
ri
b
u
t
e
d

a
d
a
p
t
a
t
i
o
n

i
n

C
ASs
W
P1
:

e
co
l
o
g
i
ca
l

mo
d
e
l
i
n
g
W
P2
:

so
f
t
w
a
re

d
i
ve
rsi
t
y


W
P3
:

d
i
st
ri
b
u
t
e
d

a
d
a
p
t
a
t
i
o
n
T

2
.
3

Exp
o
si
n
g

d
i
ve
rsi
t
y
T

2
.
2

D
i
ve
rsi
t
y
syn
t
h
e
si
s
T
2.4
D
i
ve
rsi
t
y
l
i
f
e
cycl
e

T
3.2 Goal
re
q
u
e
st

T

3
.
4

D
i
ve
rsi
t
y-
d
ri
ve
n

adaptation
T
3.3
D
i
ve
rsi
t
y
mo
n
i
t
o
ri
n
g
p
o
o
l

o
f

so
f
t
w
a
re

d
i
ve
rsi
t
y
f
f
f
f
f
f
f
f
f
f
f
f
f
f
f
T

1
.
2

D
i
ve
rsi
t
y
and
sp
e
ci
a
l
i
za
t
i
o
n
T
1.3
Ad
a
p
t
a
t
i
o
n
T

1
.
1

Eco
l
o
g
i
ca
l

b
o
a
rd

co
o
rd
i
n
a
t
i
o
n
T

4
.
1

Sce
n
a
ri
o

d
e
si
g
n

T

4
.
2

Si
mu
l
a
t
i
o
n

a
n
d

e
xp
e
ri
me
n
t

T

4
.
3

Eva
l
u
a
t
i
o
n

Contributions to
SoTA

29

co
n
t
ri
b
u
t
i
o
n
s
t
o

t
h
e

co
re

D
I
VER
SI
F
Y

o
b
j
e
ct
i
ve
D
I
VER
SI
F
Y'
s
co
n
t
ri
b
u
t
i
o
n
s
t
o

so
f
t
w
a
re

e
n
g
i
n
e
e
ri
n
g

So
T
A
so
f
t
w
a
re

d
i
ve
rsi
t
y
si
mu
l
a
t
i
o
n

f
o
r
p
e
rv
.

syst
.
mo
d
e
l
s@
ru
n
t
i
me
co
mp
o
n
e
n
t
-b
a
se
d
adaptation
d
i
st
ri
b
u
t
e
d

adaptation
Q
o
S
co
n
t
ra
ct
s
T
2.2
T
2.3
T
2.4
T
3.2
T
3.3
T
3.4
T
4.2
T
4.3
So
T
A
T
asks
W
P2
.
So
f
t
w
a
re
d
i
ve
rsi
t
y
W
P3
.
D
i
st
ri
b
u
t
e
d
adaptation
W
P4
.
C
AS
si
mu
l
a
t
i
o
n
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Background and
positioning

30

Software
diversity


The
main objective of
DIVERSIFY is
to
develop mechanisms that
introduce
diversity
at runtime, in association with the
mechanisms that select the relevant level of
diversity according to environmental
conditions.


31

Diversify & Autonomic Computing


Autonomic Computing


Adjusting the system to its
environment


How to prune the search
space?


DIVERSIFY


Adjusting the environment
to the system


Diversified search space
=>
Easy to find
a good
-
enough solution

Degree of Diversity

Probability to find solutions

Reaction time

Time needed to find a solution

Low
probability
to find a
good
-
enough
solution

Higher

probability
to find a
good
-
enough
solution

ThingML language


Modelling language for the
IoT


Based on well established formalisms


Architecture models


Asynchronous messaging


State machines


Imperative action language


Targets the whole spectrum of devices of the Future internet
(from microcontrollers to cloud)


A good candidate language for experimenting with diversity


Open
-
source and available at http://www.thingml.org

33

ThingML as a
b
ridge between
IoT

and
IoS

34

Smart City Research at TCD

Dependability, trustworthy, privacy…

Dynamic optimization of urban resources

Water
manage
-
ment

Urban
traffic
control

Community
energy
management

Smart phones

In
-
vehicle systems

CCTV

In
-
house
devices

...

...

Data brokerage and
simulation

City watch

On
-
going projects

MDDSV


Personal

Cities

LAMP

Use
case,
CityWatch

with Intel

trustworthy
participatory

& opportunistic sensing

Collecting &
disseminating
sensor data

Sensor data processing and city
environment simulation

model driven development and
formal
methods

Integrated simulation
environment
on vehicles and traffics

Formalisation of

distributed coordination

self
-
organising
of electrical
devices
on
the smart grid

Simulation on mainstream
grid simulator,
GridLAB
-
D

Multi
-
agent, single policy

DWL benchmarked

for collaborative and coordinated
smart vehicle applications

DYSARM

Runtime models to support the
adaptation of urban scale
systems

Started from the domain
analysis of water
distribution systems

Language
-
based framework
for runtime models

services
on urban
resources

Dynamic
adaptation

Runtime
models

Constraint
-
driven self
-
adaptation with user preference

Lightweight

M@R
framwork

for building
Distributed

Adaptive
System

M@R
Runtimes

for
Distributed

and
Heterogeneous

adaptive
systems


Extensible Virtual
System
Infrastructure

Resilient

Software
infrastructure
based

on
diversity

Kevoree in nutshell
1
/2

Kevoree in nutshell
2
/2

38

f
f
Refl
e
xi
ve

ca
u
sa
l

l
i
n
k
f
f
f
f
f
f
f
f
f
Legend
f
so
f
t
w
a
re

co
mp
o
n
e
n
t

xe
d

H
W

e
xe
cu
t
i
o
n

n
o
d
e
mo
b
i
l
e

H
W

e
xe
cu
t
i
o
n

n
o
d
e
co
mmu
n
i
ca
t
i
o
n

l
i
n
ks
f
f
mo
d
e
l
@
ru
n
t
i
me
co
l
l
a
b
o
ra
t
i
ve

a
d
a
p
t
i
ve

syst
e
m
f
One clonal
network

Connexions

Information
transfer
,
storage

Ramets

Resource
acquisition

The clonal plant model


More
than

70% are clonal
with

particular

network
forms


These

network
forms

are
constitued

of
two

units

with

different

functions

heterogeneous

environments

Ramet
specialization

(diversification of
functions
)

Low

resource

environments

The clonal plant model


In
heterogeneous

environments
, apparition
of
diversity

within

the
network


Importance of
heterogeneity

grain,
environment

predictability
, patch
contrasts


Scale

of signal
integration


Treshold

for
response

development

(
trade
-
off
cost

vs.
theoretical

benefit

at

the network
level
)

Local
environmental

cues

(change in
environmental

conditions,
stress, local
disturbance
)

Local
response

(
growth
, (reproduction))

Global network
performance

(
efficiency

in
resource

acquisition (
-
>
biomass
), network
survival
)

The clonal plant model


Two

way

for
diversity

development
:


spontaneous

(
age

dependent
)


response

to
environmental

changes

Relation
with

close
projects

42

Relation with
PerAda

and Awareness


Common


Focused on the self
-
awareness and self
-
adaptation of software
-
based systems


Started from large data, services, and learning
-
based technologies


Share some common topics (such as e
-
Mobility with ACENS, a
project under Awareness)


Difference


We focus more on
the urban
infrastructure (water, energy), rather
than the social aspects


We focus more on software (services), rather than control (robots)


We are from a software engineering perspective, utilizing MDE,
middleware technology, etc.


43

Related projects
-

AWARENESS
1/
5


Sapere = Self
-
aware Pervasive Service Ecosystems

o
Model and deploy services as autonomous individuals in an ecosystem of
other services, data sources, and pervasive devices.


Self
-
aware components and a general nature
-
inspired interaction model


Decentralized self
-
* algorithms


Spatial self
-
organization, self
-
composition, and self
-
management


Diversify will takes inspiration from Ecology by
involving Ecologists

in the
project, and will mainly focus on leveraging the
diversity and food web
properties

from Ecosystems to build reliable
systems

Related projects
-

AWARENESS 2/
5


Cocoro =
Collective Cognitive Robots

o
Swarm intelligence inspire from natural and biology phenomenon


Application to: robots, underwater vehicles ...


Diversify will not focus on this type problems.


45

Related projects
-

AWARENESS
3/5


ASCENS
: Autonomic Service
-
Component
ensemble
s


Combine formal method and optimal resource usage promised by autonomic
computing


Apply to robotics, cloud computing and
e
-
Vehicles


Diversify
has a totally different approach build self
-
adaptive resilient
applications inspired by
eco
-
systems


EPICS
: Engineering Proprioception in Computing
Systems


Proprioception (coming from psychology) is the basic ability to collect and
maintain information about state and progress


Transfer knowledge from another science to computer
science


Diversify
follow the same process for transferring knowledge from
another science to computer science, but will focus on transferring
knowledge from Ecology and will integrate Ecologist as core partners
of the project

46

Related projects
-

AWARENESS
4/5


Recognition: Relevance and cognition for
self
-
awareness in a content
-
centric Internet

o
inspired by the cognitive process of human

o
using psychological and cognitive science


apply to Internet content


Diversify is not doing the same thing ...

Related projects
-

AWARENESS
5/5


SYMBRION : Symbiotic Evolutionary Robot
Organisms

o
swarm & collective robot systems
-

evolutionary
robot organisms


apply to flots of robots


Symbrion is cited in PerADA and Awareness


A bit particular because the website speaks about 3
projects (one in Awareness and the other in PerADA) +
Symbrion Enlarged EU + un projet REPLICATOR


Related Project
-

PerAda 1/5


ALLOW : Adaptable Pervasive Flows

o
New programming paradigm for developing
adaptable pervasive flows


Compared to Diversify : Use traditionnal
techniques of Context
-
aware programming
and so on.


Related Project
-

PerAda 2/5


ATRACO : Adaptive and Trusted Ambient
Ecologies

o
A context
-
aware artefact, appliance or device
uses
sensors to perceive

its
context

of operation and
applies an ontology to interpret this context. It also
uses internal
trust models

and fuzzy decision
making mechanisms to
adapt

its operation to
changing context.


Diversify will works with Ecologist to really
transfer knowledge from Ecology to
computer science.

Related Project
-

PerAda 3/5


FRONTS : Foundations of Adaptive
Networked Societies of Tiny Artefacts

o
foundational algorithmic

o
unifying scientific framework and a coherent set of
design rules
, for global systems resulting from the
integration

of
autonomous interacting entities
,
dynamic
multi
-
agent

environments and
ad
-
hoc
mobile networks
.

Related Project
-

PerAda 4/5


REFLECT
: Responsive Flexible
Collaborating Ambient

o
sensing users and their
mood
and
intentions
+
human behavioural patterns

=
environmental
awareness

==> used for adaptation


Very different to what we are doing in
Diversify

Related Project
-

PerAda 5/5


SOCIALNETS: Social networking for
pervasive adaptation

o
how social networks can be exploited for the
delivery and acquisition of content, including issues
of security and trust


Very different to what we are doing in
Diversify