Adaptive Systems Lecture 10: Engineering Adaptive Systems

1
Adaptive SystemsLecture 10 Engineering
Adaptive Systems

• Dr Giovanna Di Marzo Serugendo
• Department of Computer Science
• and Information Systems
• Birkbeck College, University of London
• Email dimarzo_at_dcs.bbk.ac.uk
• Web Page http//www.dcs.bbk.ac.uk/dimarzo

2
Lectures 8/9 Review

• Taxonomy / Classification
• Optimisation Problems (Static/Dynamic)
• Cultural Heritage (Bees)
• Trust-based access control
• Swarms of Robot (Ants)
• Region detection (Spiders)
• Manufacturing Control (Ants)
• Immune Computer
• P2P Systems Protocols (Gossip)
• Autonomic Computing

3
Lecture 10 Overview

• Overall picture
• Software Agents
• Autonomous Agents
• Mobile Agents
• Adaptation Mechanisms
• Translation of natural mechanisms
• Artificial mechanisms
• Middleware infrastructures
• Swarm-based
• Field-based (Lecture 11)
• Methodologies and tools
• Open Issues

4
Overall Picture
Adaptation Mechanisms
Natural Adaptive Systems
Engineering Adaptive Systems
Model State st
Artificial Adaptive Systems
Analysis and Simulation
5
Engineering Overview
Engineering Adaptive Systems
6
Software Agents

• Definition Wooldridge 01
• An agent is a computer system situated in some
  environment, and that is capable of autonomous
  action in this environment in order to meet its
  design objectives
• Characteristics
• Executing program
• Acts on behalf of a user or a program
• Autonomous
• Proactive (take initiative)
• Distributed
• Maintain information about own resources
• Intelligent
• Mobile

7
Software Agents

• Agents
• Situated in an environment
• Sensors (to sense environments state)
• Actuators (to perform some action based on
  environments state and own tasks)
• Reactive Agents
• Act in response to some event/change occurring in
  the environment
• Autonomous Agents
• Make independent decisions
• Take initiatives
• Intelligent Agents
• Adapt and learn
• Reasoning (Belief-Desire-Intention BDI)
• Mobile Agents
• Move from one computer to another one

8
Software Agents

• Agents Interactions
• Communication (exchange of messages)
• Inform, request, confirm, etc.
• Coordination (of individual agents tasks)
• Managing inter-dependencies between the
  activities of the agents
• Resource sharing
• Dependencies (tasks scheduling)
• Collaboration
• Competition
• Transactions
• Agent-oriented software engineering

9
Multi-Agent Systems

• Multi-Agent Systems
• Systems composed of several agents
• Reaching goals difficult to achieve for single
  agent
• May
• include humans
• exhibit self-organising behaviour

10
Multi-Agent Systems

• Applications
• Distributed Systems
• Multi-agent system carries out some task / Nodes
  of systems agents
• Distributed sensing
• Vehicles monitoring using distributed cooperating
  sensors
• Information retrieval and management (web)
• Control air-traffic control, spacecraft control
• Personal Software Assistants
• Individual agents
• Proactive assistant helping user with some
  application
• Calendar assistant, e-mail assistant
• E-commerce
• Shopping agents, auction bots
• Virtual environments

11
Multi-Agent Systems

• Applications
• Simulation and Modelling
• Social systems (human societies)
• Complex systems

12
Mobile Agents

• Software Agents
• Move from one machine to another
• Types of Mobility
• Strong Mobility
• Code execution stack are moved
• Execution continues where it stopped
• Weak Mobility
• Code only is moved
• Possibly data
• Execution starts at the beginning

13
Mobile Agents

• Elements of mobility
• Components
• Resources
• Source Code
• Data
• Hardware (CPU)
• Calculus
• Process, threads execution state, links
• Execution
• Interactions
• Message exchange
• Execution sites (hosts)

14
Mobile Agents

• Execution of a service
• Requirements
• Know-how (code)
• Data
• Physical resources (CPU)
• Process
• Need to be at the same place

15
Mobile Agents

• Mobility Paradigms Carzaniga 97, Fuggetta 98
• Client/Server
• Remote Evaluation
• Code on Demand
• Mobile Agent

16
Mobility Paradigms

• Client/Server

17
Mobility Paradigms

• Remote Evaluation

18
Mobility Paradigms

• Code on Demand

19
Mobility Paradigms

• Mobile Agents

20
Mobile Agents

• Technology
• Programming Languages
• Run-time
• Execution support
• Platforms/Virtual Machines

21
Mobile Agents

• Run-time environment
• Mobility primitives
• Submit / clone / etc
• Interaction Mechanisms
• Agents coordination
• Type of mobility
• Programming
• Aware of execution site
• Mobility is under programmers control

22
Mobile Agents

• Applications
• History
• Load balancing in OS
• Remote batch processes
• Postscript files
• Independent initiatives for building virtual
  machines
• Java
• Applets, Aglets (Java mobile code)
• Active networks
• programs in network nodes

23
Adaptation Mechanisms

• Translation of Natural Mechanisms
• Stigmergy
• Indirect communication through the environment
• Digital pheromone
• intensity, evaporation rate
• Work-in-progress
• mapping table configurations action
• Gossiping
• Informal discussion among entities
• Local exchange of information (neighbours list)

24
Adaptation Mechanisms

• Translation of Natural Mechanisms
• Trust
• Human trust in peers
• Trust values, calculation of risk, decision of
  actions
• Updated on basis of positive/negative evidence
• Immune System
• B cells antibodies (detection and marking)T
  cells (destruction)
• Bit strings (anomaly to detect)
• Mobile agents (B and T cells)

25
Adaptive Mechanisms

• Implementation of Artificial Mechanisms
• Tags
• Markings attached to individuals (agents) and
  observable by others
• Agents change behaviour on basis of utility
  function value observed in peers (tag)
• Metadata
• Additional information (metadata) about
  functional / non-functional information /
  policies
• Middleware processes metadata and components
  adapt to policies

26
Adaptive Mechanisms

• Middleware infrastructure
• Supporting adaptive mechanism
• Built-in adaptation features

27
Middleware Infrastructures

• Middleware
• Intermediary software layer
• Allow communication and coordination among agents
  (entities)
• Interest for decentralised adaptive systems
• Shared environment
• Agents coordinate their work through this
  environment
• No need for central control
• May support adaptation mechanism

28
Middleware Infrastructures

• Coordination Spaces
• Blackboard
• Repository of tuples (n-uples) accessed
  concurrently
• Producers and consumers of tuples
• Associative memory
• Retrieval of tuples based on pattern matching
• Swarm-Based Infrastructures
• SwarmLinda
• AntHill
• Field-Based Infrastructures
• Co-Fields
• TOTA

29
Coordination Spaces

• Linda Gelernter 85
• Coordination language/model based on shared tuple
  spaces
• Indirect communication
• Insertion of tuples in the shared data space
  (out)
• Retrieval of tuples from the shared data space
  (in or rd)
• Retrieval is based on matching a given template

30
Coordination Spaces
Shared Tuple Space
A
(name, Arthur, surname, Smith, age, 7)
rd
B
(name,?x, surname, Smith, age, ?i)
31
Coordination Spaces

• Coordination spaces as middleware layers
• Uncoupled interactions
• Limited form of self-organisation
• Decentralised control, anonymous and indirect
  interactions among agents
• No specific support for adaptation mechanism
• Languages
• Linda
• JavaSpaces (Sun)
• TSpaces (IBM)

32
SwarmLinda

• SwarmLinda System Charles 04
• Network of tuple space nodes
• Nodes communicate with each other
• Node exchange tuples
• Goal
• Optimisation distribution and retrieval of tuples
• Dynamically determining storage locations for
  tuples
• Tuples of same type stored in clusters (clusters
  emerge)
• Interest
• makes Linda scalable and distributed

33
SwarmLinda

• Principle Ant Metaphor
• Ants
• Template-ant search for food (matching tuple)
• Tuple-ant (store food store tuples)
• Ants environment
• Network of tuple spaces nodes
• Content tuples and scent (likelihood of matching
  templates)
• Scent stands for the pheromone
• Tuple Storage (out)
• Storage ant sorting (Tuple-ant)
• Tuple-ants carry tuple and drop tuple at specific
  nodes
• Tuple Retrieval (in / rd)
• Requests ants looking for food (Template-ant)
• Template-ants carry request and test at each node
  for matching tuples
• http//cs.fit.edu/rmenezes/SwarmLinda

34
SwarmLinda

• Implementation
• Scent is volatile and disappears with time
• Storage
• Node N augments  scent  for the same kind of
  tuple
• If sufficient  scent  then keeps the tuple
• If not, looks for more suitable neighbour
  (highest concentration of that scent).
• Tuple is sent to that node (that will do the same
  with the tuple keep or send elsewhere).
• Tuple is dropped after some time even if no
  matching scent
• Retrieval
• Node N determines if a local tuple matches the
  template
• If no matching tuple looks for highest scent for
  that kind of tuple in its neighbourhood.
• Request is sent to that node (that will do the
  same with the request)

35
SwarmLinda

• Algorithms
• Retrieval
• Initial spreading of scent close to nest
• Loop
• Template-ant checks for matching tuple at current
  location
• Found - go back at origin - leave scent
  at each step
• Not found - Check neighbour for highest scent
  and move there - If no scent, choose randomly
  a neighbour

36
SwarmLinda
37
Anthill

• Anthill System Babaoglu 02
• Dynamic network of peer nodes
• Adaptive agents travel through the network to
  solve complex problems
• Interact with nodes
• Cooperate with other agents
• P2P Applications

38
Anthill

• Principle - Ant metaphor
• Anthill system network of interconnected nests
• Nest peer entity sharing computational and
  storage resources
• Nests handle requests of local applications
  (user)
• Generate one or more ants roaming the network to
  satisfy the request
• Ants
• Observe environment
• Perform local computation
• Move from nest to nest until finds information
• Leave information on visited nests (e.g. routing
  info for retrieving files)
• Indirectly communicate with each other
  (stigmergy)
• Service for applications
• Implemented through the ants work
• E.g. file-sharing application / distributed file
  index search service
• Applications interactions with Services
• Application perform service requests to local
  node
• Wait for replies through local node

39
Anthill
File Sharing Service
Files Storage
Grid Service
Data Storage
40
Readings

• Babaoglu 02 Özalp Babaoglu, Hein Meling,
  Alberto Montresor Anthill A Framework for the
  Development of Agent-Based Peer-to-Peer Systems.
  ICDCS 2002 15-22
• Carzaniga97 A. Carzaniga, G.-P. Picco, G.
  Vigna Designing Distributed Applications with
  Mobile Code Paradigms. International Conference
  on Software Engineering (ICSE97), pp. 22 - 32,
  1997.
• Charles 04 A. Charles, R. Menezes, R.
  Tolksdorf. On the Implementation of SwarmLinda
  A Linda System Based on Swarm Intelligence.
  CS-2004-03. Florida Tech, Computer Science, 2004.
• Fuggetta98 Alfonso Fuggetta, Gian Pietro Picco,
  Giovanni Vigna. Understanding Code Mobility.
  IEEE Transactions on Software Engineering 24(5),
  pp. 342 361, 1998.
• Gelernter 85 D. Gelernter. Generative
  communication in Linda. ACM Transactions on
  Programming Languages and Systems, 7(1)80--112,
  January 1985.
• Wooldridge 01 M. Wooldridge. An Introduction
  to Multi-Agent Systems. Wiley. 2001.