ON SELF-ADAPTATION
IN SYSTEMS-OF-SYSTEMS
M. Pilar Romay, UAH & UCM, Spain
Carlos E. Cuesta, VorTIC3, URJC, Spain
Luis Fern...
CONTENTS
 Introduction
 Towards Adaptive SoS
 Self-Adaptation to Architecture
 Dynamic & Adaptive Architecture
 Forma...
INTRODUCTION
 SoS is an structural concept
 Defined by their composite emergent behavior
 Depending on the architecture...
TOWARDS ADAPTIVE SOS (I):
SELF-ADAPTATION TO ARCHITECTURE
 Self-adaptation research has many facets
 Emergent and autono...
TOWARDS ADAPTIVE SOS (II):
DYNAMIC & ADAPTIVE ARCHITECTURE
 Definition of Dynamic ADLs
 Complex architectures (including...
TOWARDS ADAPTIVE SOS (III):
ADAPTATION IN SOS
 Facets of adaptation in the SoS definition
 Autonomy, connectivity and em...
ARCHITECTING ADAPTIVE SOS (I):
THE CASE OF LOCI
7
 Structuring architecture in
terms of evolution
 Modules defined as “a...
ARCHITECTING ADAPTIVE SOS (II):
THE CASE OF STRATA
8
 The rhythm of change is
different at different scales
 Software Ev...
STUFF
SERVICES
SKIN
STRUCTURE
SITE
SPACE PLAN
ARCHITECTING ADAPTIVE SOS (III):
PACE LAYERING
9
 Deriving from “building”
...
CONCLUSIONS
 There is a deep relationship between adaptivity and
systems-of-systems
 Research in self-adaptive (and auto...
THANKS FOR YOUR ATTENTION
11
Upcoming SlideShare
Loading in …5
×

SESoS 2013 (Romay)

355 views
192 views

Published on

Published in: News & Politics, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
355
On SlideShare
0
From Embeds
0
Number of Embeds
6
Actions
Shares
0
Downloads
3
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

SESoS 2013 (Romay)

  1. 1. ON SELF-ADAPTATION IN SYSTEMS-OF-SYSTEMS M. Pilar Romay, UAH & UCM, Spain Carlos E. Cuesta, VorTIC3, URJC, Spain Luis Fernández-Sanz, UAH, Spain Montpellier, France, 02/07/2013
  2. 2. CONTENTS  Introduction  Towards Adaptive SoS  Self-Adaptation to Architecture  Dynamic & Adaptive Architecture  Formalizing Adaptive Architecture  Adaptation in SoS  Architecting Adaptive SoS  Change: the notion of Locus  Scale: the notion of Stratum  Pace Layering  Conclusions & Future work 2
  3. 3. INTRODUCTION  SoS is an structural concept  Defined by their composite emergent behavior  Depending on the architecture of the SoS  Adaptivity as an emergent behavior  It is also one of the defining features of a SoS  Operational independence, Emergent behavior (Maier)  Autonomy, connectivity, emergence (Boardman & Sauser)  Self-adaptation encompasses many facets  Includes aspects such as autonomy and emergence  This relationship must be exploited  Adaptive Architecture as the basis of SoS  A dynamic architecture of adaptive services 3
  4. 4. TOWARDS ADAPTIVE SOS (I): SELF-ADAPTATION TO ARCHITECTURE  Self-adaptation research has many facets  Emergent and autonomous self-adaptation  Autonomic Systems  Internal operation without external assistance  Self-organizing Systems & Architectures  Adaptive architectures  Origin in self-* Systems  Dynamic software architecture  Including self-configuration  Evolving into self-adaptive system architecture  Self-healing, dependability  Adaptivity as a generic notion 4
  5. 5. TOWARDS ADAPTIVE SOS (II): DYNAMIC & ADAPTIVE ARCHITECTURE  Definition of Dynamic ADLs  Complex architectures (including SoS)  Formal approaches (esp. -calculus-based)  Self-adaptive architectures as their evolution  The next step in Sw Eng (Kramer & Magee)  Approaches to adaptive architecture  General-purpose middleware, e.g. Rainbow  Domain-specific middleware, e.g. Music  Synchronized, reflective, policy-based architectures  Difficult to measure “self-attributes”  Lack of a clear reference model  A formal approach (process calculus) is advocated 5
  6. 6. TOWARDS ADAPTIVE SOS (III): ADAPTATION IN SOS  Facets of adaptation in the SoS definition  Autonomy, connectivity and emergence  SoS requires that “connective media” are autonomic  Self-adaptive architecture model of emergence  (Still) Lack of a high-level architecture approach  Preliminary: exporting work from other contexts  Service-oriented, Model-driven, Dynamic Arch.  Promising approach: Federated Systems  Adaptive architecture as a good basis  High-level patterns must still be generalized  A formal foundation seems necessary  Measuring capabilities would be also required 6
  7. 7. ARCHITECTING ADAPTIVE SOS (I): THE CASE OF LOCI 7  Structuring architecture in terms of evolution  Modules defined as “areas of change” rather than functionality  Locus (Morrison, 2007)  Change contexts: parts of the system which always evolve in synchrony  Evolutionary steps  Incarnations of a locus  Reminiscent of Evolution Styles (more recent) Composition of Loci, using the Evolver-Producer pattern (Morrison et al, 2007)
  8. 8. ARCHITECTING ADAPTIVE SOS (II): THE CASE OF STRATA 8  The rhythm of change is different at different scales  Software Evolution  Software is able to follow different evolutionary patterns at different levels  Known as strata  Including SoS  Lower strata evolve easily  Upper strata evolve is much less frequent  Upper strata change often imply structural change Strata of Evolution, according to (Mittermeir, 2006) System-of-Systems System Architecture Design Unit
  9. 9. STUFF SERVICES SKIN STRUCTURE SITE SPACE PLAN ARCHITECTING ADAPTIVE SOS (III): PACE LAYERING 9  Deriving from “building” architecture theory  Shearing Layers of Change  Processes affect systems in different timescales  Different parts (layers) are evolving at a different pace  Able to adapt = slippage of layers  Design principle: structure layers according to this  Seems natural in SoS  Administrative barriers Shearing Layers of Change, according to (Brand, 1994)
  10. 10. CONCLUSIONS  There is a deep relationship between adaptivity and systems-of-systems  Research in self-adaptive (and autonomic) systems can (must) be applied in this context  Adaptive architecture seems to be a reasonable approach to tackle their design  System evolution itself appears as the main driver for adaptation of SoS  Defined as the highest stratum  Pace layering as a reasonable design strategy  Just a first step in this direction 10
  11. 11. THANKS FOR YOUR ATTENTION 11

×