Description and Composition of Bio-Inspired Design Patterns: The Gradient Case

Description and Composition of
Bio-Inspired Design Patterns:
The Gradient Case
1
Jose Luis Fernandez-Marquez
University of Geneva, Switzerland
Joseluis.fernandez@unige.ch
http://iss.unige.ch
In collaboration with:
Giovanna Di Marzo Serugendo – University of Geneva, Switzerland
Josep Lluis Arcos – IIIA-CSIC, Barcelona, Spain
Mirko Viroli - University of Bologna, Italy
Sara Montagna - University of Bologna, Italy

Outline
  Motivation
  Goal
  Bio-Inspired Design Patterns
  Gradient Pattern
  Chemotaxis Pattern
  Applications
  Framework: SAPERE Project
  Conclusions and Future research
2

Motivation
3
  Characterized by:
  Large Scale
  Openness
  Unpredictability
  Wide range of new
applications
  Requirements:
  Scalability
  Robustness
Traditional Approaches
(centralised, not distributed)

Motivation
4
  Bio-Inspired Self-Organising
mechanisms have been applied in those
infrastructures, achieving results that go
beyond traditional approaches, (ACO,
PSO, flocking, Digital pheromones….) .
However,
  The knowledge and experience on how,
when, and where to use them is spread
across the corresponding literature.
  It is very difficult to grasp what are their
capabilities and weakness.

Goal
5
  To analyse existing literature, providing a catalogue of
Bio-inspired Mechanisms for Self-Organizing Systems.
  To describe those mechanisms as design patterns,
identifying how, where, and when to be applied.
  Identify the relationship between the presented
mechanisms, providing a better description and making it
easier to compose new patterns or adapt the existing
patterns to solve new problems.
  Demonstrate the applicability of those mechanisms
tackling with different domains:
  Dynamic Optimization
  Spatial Computing
  Sensor Networks

Bio-Inspired Design
Patterns
8
Repulsion Evaporation Replication Aggregation Spreading
Flocking Foraging Chemotaxis Morphogenesis
Quorum Sensing
Digital Pheromones Gradients Gossip

Bio-Inspired Design
Patterns
9
How
When Where
Aliases Biological Inspiration Related Patterns
Typical Case
Known Uses
Problem
Environment
Solution
Forces
Entities / Dynamics
Implementation
Consequences
Pattern Description

The Gradient Pattern
10
  Problem: Large systems suffer from lack of global knowledge to estimate
the consequences of the actions performed by other agents beyond their
communication range.
  Solution: Information spreads from the location it is initially deposited and
aggregates when it meets other information. During spreading, additional
information about the sender's distance and direction is provided: either
through a distance value (incremented or decremented); or by modifying
the information to represent its concentration (lower concentration when
information is further away).
  Abstract transition rule:

11
  Dynamics:

12
  Dynamics:

The Chemotaxis
Pattern
13
  Problem: Decentralised motion
coordination aiming at detecting sources or
boundaries of events.
  Solution: Agents locally sense gradient
information and follow the gradient in a
specified direction (either follow higher
gradient values, lower gradient values, or
equipotential lines of gradients).
  Abstract transition rule:

The Chemotaxis
Pattern
14
  Dynamics:

Applications
15
  Dynamic Optimisation:
  We extended PSO with the Evaporation Pattern to deal with
dynamic and noisy optimisation.
  Hovering Information in Spatial Computing:
  We defined and analysed a collection of algorithms based on
the Replication Pattern and the Repulsion Pattern, for
persistent storage of information at specific geographical
areas.
  Detecting Diffuse Events Sources
  We implemented the Chemotaxis Pattern for localizing
dynamically changing diffuse events using WSN.

FACULTÉ DES SCIENCES
ÉCONOMIQUES ET SOCIALES
Département des Hautes Etudes Commerciales -HEC
Framework:
SAPERE project
  Theoretical and practical framework for
decentralized development and execution of
self-aware and adaptive services for future and
emerging pervasive network scenarios.
  Chemical Interactions among Services
  Smooth data/service distinction
  Spontaneous interactions of
available services
  Bio-chemical reactions
  Middleware for Android phones / tablets
  Context-awareness (user, situation recognition)
  Case Study
  Focus on public/private displays for crowd steering
  Domains
  Context-Aware Advertisement, Crowd Steering, User guidance
  EU Funded Project (SAPERE: http://www.sapere-project.eu)
  Collaboration: U Geneva, U Bologna, U Modena, U Linz, U St-Andrews
  2010-2013

FACULTÉ DES SCIENCES
ÉCONOMIQUES ET SOCIALES
Département des Hautes Etudes Commerciales -HEC
Framework:
SAPERE project
Crowd Steering through Self-Organising
Public Displays
•  Collaborative displays
•  Self-organising spontaneous interactions
  Bio-inspired (gradients, gossip,
stigmergy, flocking)

Conclusions
18
  This work is a step forward for engineering self-Organising
Systems.
  We presented a catalogue of bio-inspired Self-Organising
mechanisms, as design patterns.
  We analysed the relations between the mechanisms,
making easier their composition and adaptation to solve
new problems.
  We contributed in different domains using bio-inspired
Self-Organising Mechanisms:
  Dynamic Optimisation (Evaporation mechanism)
  Sensor Networks (Chemotaxis mechanism)
  Spatial Computing (Replication + repulsion)

Future Works
19
  SAPERE Project:
  Add New Patterns in the catalogue.
  Self-Adaptation of parameters.
  Self-Composition of patterns.
  Implementation of services.

Any questions?
Thank you for your attention!
20
Jose Luis Fernandez-Marquez
Joseluis.fernandez@unige.ch

Description and Composition of Bio-Inspired Design Patterns: The Gradient Case

More Related Content

Similar to Description and Composition of Bio-Inspired Design Patterns: The Gradient Case (20)

Recently uploaded (20)

Description and Composition of Bio-Inspired Design Patterns: The Gradient Case

Editor's Notes