Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 53

Pervasive Computing

BITS Pilani Gokul Kannan Sadasivam


Department of CS & IS
Hyderabad Campus
Overview

• Intelligent System Interaction


• Introduction
• Cooperative Interaction
• Competitive Interaction

• IS Design Principles
• Intelligent Interaction Protocols
• Multi-Agent Systems

11/12/2022 BITS SS ZG531 (Pervasive Computing) 2 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Introduction

3 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
Intelligent (Smart) interaction

What is Intelligent (Smart) interaction?


• Beyond using universal network communication
protocols
• Involves Coordination, Competition, etc.

2 dimensions of intelligent interaction


• Interaction between multiple intelligent systems & their
environments
• Intelligent Interaction between relatively non-intelligent
multiple systems & environments

11/12/2022 BITS SS ZG531 (Pervasive Computing) 4 of 55

BITS, Pilani - Hyderabad Campus


Nature of Communication

• Hidden from applications (lower-level or middleware


services)
• Variable paths
• Variable delays
• Variable ordering
• Variable attenuation
• Time synchronization
• Space synchronization

11/12/2022 BITS SS ZG531 (Pervasive Computing) 5 of 55

BITS, Pilani - Hyderabad Campus


P2P Interaction Between Multiple
Senders and Receivers

• MIMO systems - Multiple Input Multiple Output systems

• MAS - Multi-Agent Systems

11/12/2022 BITS SS ZG531 (Pervasive Computing) 6 of 55

BITS, Pilani - Hyderabad Campus


Interaction Multiplicity
Examples: Communication

11/12/2022 BITS SS ZG531 (Pervasive Computing) 7 of 55

BITS, Pilani - Hyderabad Campus


Unknown Sender and
Malicious Senders
• Sender masquerades
• Unknown senders' messages should be tested in
quarantined area
• E.g. Sensor Networks

11/12/2022 BITS SS ZG531 (Pervasive Computing) 8 of 55

BITS, Pilani - Hyderabad Campus


Unknown Receivers

• Multicast (or Broadcast) message transmission


• Sender interacts with an unknown receiver:
a) to share information
b) to request a response
• Used in service discovery in dynamic open systems
• Used in p2p interaction

11/12/2022 BITS SS ZG531 (Pervasive Computing) 9 of 55

BITS, Pilani - Hyderabad Campus


Too many messages

• Problems
• Receivers may become over-loaded
• Big issue in synchronous communication

• Possible Solutions
• Policy-based management - priority based handling events, FCFS, etc.
• Buffers can be used
• Receivers can support flow-control
• Proxies can off-load event processing
• Filtering of events

11/12/2022 BITS SS ZG531 (Pervasive Computing) 10 of 55

BITS, Pilani - Hyderabad Campus


Mediated Interaction

• Mediator: between interacting participants / peers


– also referred to as third-parties, intermediaries, middle-ware agents

• Advantages of Mediators
• Heterogeneous
• Reduces complexity of interaction

• Some disadvantages
• Performance
• Single point of failure
• Man-in-the-middle attack

11/12/2022 BITS SS ZG531 (Pervasive Computing) 11 of 55

BITS, Pilani - Hyderabad Campus


Shared Communication
Resource Access
• Need transmission control or co-operation

• Hidden-node problem

• Use of multiplexing is encouraged - TDMA, FDMA, CDMA

• Collision Avoidance schemes

• Collision Detection schemes


a) Used in applications like simulated games
b) Used in robots to detect obstacles

11/12/2022 BITS SS ZG531 (Pervasive Computing) 12 of 55

BITS, Pilani - Hyderabad Campus


Shared Computation
Resource Access
• Sharing of resources for interaction
• Multiple access control schemes
• Concurrency control on data (RDBMS database)
• Token ring network to access resources
• MTOS to run multiple applications in a single device

11/12/2022 BITS SS ZG531 (Pervasive Computing) 13 of 55

BITS, Pilani - Hyderabad Campus


Mediating Between
Requestors and Providers
Mediators
• Provides loose coupling
• Enhances peer discovery and service discovery
• Service capability - what type of services can be
obtained from the provider
• Service preferences
• requesters' constraints when selecting a service provider
• constraints can be kept public or private or a mixture of both

11/12/2022 BITS SS ZG531 (Pervasive Computing) 14 of 55

BITS, Pilani - Hyderabad Campus


Mediating Between
Requestors and Providers
Who Knows Capabilities
Who Knows Requests & Preferences

11/12/2022 BITS SS ZG531 (Pervasive Computing) 15 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Cooperative Interaction

16 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
Cooperative Interaction

• Characterised by the following properties


• Coordination: synchronising activities
• Cohesion: acting together

• Easy to manage when


• Homogeneous systems
• Centralized control
• Systems are mandated to serve others

• Cooperation is harder to manage when


• Different systems are designed by independent developers
• Systems are autonomous
• Systems support heterogeneous goals
• Parties work in self-interested manner

11/12/2022 BITS SS ZG531 (Pervasive Computing) 17 of 55

BITS, Pilani - Hyderabad Campus


Coordination

• Cooperators share their goals with collaborating parties


& act together.

• Need for coordination


• Dependencies between agents’ actions

• IS knew about its local actions and the (anticipated)


actions of others to ensure that the community acts in a
coherent manner.

11/12/2022 BITS SS ZG531 (Pervasive Computing) 18 of 55

BITS, Pilani - Hyderabad Campus


Perfect Coordination

• Each system has a good model of other systems (Mutual


Modelling).
• No need for explicit communication.

11/12/2022 BITS SS ZG531 (Pervasive Computing) 19 of 55

BITS, Pilani - Hyderabad Campus


Coordination Classification

• Explicit coordination
• Implicit coordination

11/12/2022 BITS SS ZG531 (Pervasive Computing) 20 of 55

BITS, Pilani - Hyderabad Campus


Explicit coordination

• Cooperative Distributed Problem Solving – CDPS


• Type of distributed planning
• Decomposing problems into sub-problems
• Allocating sub-problems to different participants
• Exchanging sub-problem solutions
• Creating the overall solution out of all the sub-problem solutions

11/12/2022 BITS SS ZG531 (Pervasive Computing) 21 of 55

BITS, Pilani - Hyderabad Campus


Implicit coordination

• Norms and Electronic Institutions


• Norms: a set of standards.
• Normative behaviour: used to check if actual behaviour is legal or illegal.
• Institutions are normative systems built upon agreed constraints on individual
behaviour.
• E.g.: Passenger vehicle first allow to leave and then pick up.

• Hierarchical & Role-based Organisational Interaction


• What is an organisation? (Roles, Missions)
• Hierarchical Organisation: DNS, OO class hierarchies

11/12/2022 BITS SS ZG531 (Pervasive Computing) 22 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Competitive Interaction

23 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
Competitive Interaction

• Also, called as Adversarial Interaction

• Interaction that is driven by self-interests

• Focus on autonomous participants furthering their own goals (rather


than other or group goals)

• E.g. Utility regulation scenario


• Multiple autonomous lighting devices in a room.
• All turning on at the same time, when the user enters.
• Waste of energy.

• Challenges in competitive interaction designs


• Some systems may act maliciously
• Self-interested agents keep their goals private

11/12/2022 BITS SS ZG531 (Pervasive Computing) 24 of 55

BITS, Pilani - Hyderabad Campus


Market-based Interaction

• Used to allocate resources to individual requestors

• Two principles of micro-economics


a) Optimization - the participants choose the best deal that maximizes their payoff
b) Equilibrium - demand equals supply

• Each provider and consumer is self-interested

11/12/2022 BITS SS ZG531 (Pervasive Computing) 25 of 55

BITS, Pilani - Hyderabad Campus


Auctions

• Sellers versus Bidders

• Designed to allocate resources to one of the bidders.

• Types of auction protocol depends on


• Properties for how bids can be made
• Dissemination policy (what is revealed to whom and when)
• Clearing policy

11/12/2022 BITS SS ZG531 (Pervasive Computing) 26 of 55

BITS, Pilani - Hyderabad Campus


Negotiation

• Three principle components


1) Exchange of proposals and counter-proposals
2) Deal is reached
3) Negotiation set (a set of all agreements)

• Modelled using game theory

• Argumentation-based negotiation
• Giving additional information about negotiation

• Applications
• Used in auctions
• Used in reverse auctions

11/12/2022 BITS SS ZG531 (Pervasive Computing) 27 of 55

BITS, Pilani - Hyderabad Campus


Agreements

• Needed for peers to synchronize actions


• Formal or informal
• Two/multiple participants
• Convergence
• a set of protocols and algorithms to reach agreements
• e.g. router table updates

11/12/2022 BITS SS ZG531 (Pervasive Computing) 28 of 55

BITS, Pilani - Hyderabad Campus


Consensus-based
Agreements
• Multiple participants
• A common goal
• Plan is private
• Voting
• Registration – compiling a list of participants eligible to vote
• Validation – Checking the credentials of the participants
• Casting – Voters making their selection
• Collection
• Tallying – counting the votes

11/12/2022 BITS SS ZG531 (Pervasive Computing) 29 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Is Interaction Design

30 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
IS Interaction Design

Two basic dimensions to supporting intelligent interaction:


• to design conventional system interaction to be
intelligent
• to design individual intelligent systems to interact

11/12/2022 BITS SS ZG531 (Pervasive Computing) 31 of 55

BITS, Pilani - Hyderabad Campus


More Intelligent Conventional
system Interaction
Motivation
• Mediation & handling heterogeneity
• Reflection about communication
• Distributed problem solving
• Task delegation
• Flexibility and Selection
• Reliability

11/12/2022 BITS SS ZG531 (Pervasive Computing) 32 of 55

BITS, Pilani - Hyderabad Campus


Interaction Between Individual
ISs: Motivation
• Knowledge Bootstrapping problem
• Individual IS has to learn everything on its own.
• Need a complete knowledge model of the World

11/12/2022 BITS SS ZG531 (Pervasive Computing) 33 of 55

BITS, Pilani - Hyderabad Campus


Interaction Protocol Types

Individual messages are not used in isolation but used in


different patterns of multiple messages

Classification of Interaction Protocols


• Information sharing vs. task sharing
• Unicast versus Multicast
• Pull versus Push
• Syntactic versus Semantic

11/12/2022 BITS SS ZG531 (Pervasive Computing) 34 of 55

BITS, Pilani - Hyderabad Campus


Interaction Protocol Design

• Message: header + body

• Message header:
• message type
• the receiver address
• the sender address
• the message content encoding
• timestamp, etc.

• Types of messages
• HTTP/1.1 - Request, Response

11/12/2022 BITS SS ZG531 (Pervasive Computing) 35 of 55

BITS, Pilani - Hyderabad Campus


Message patterns

a) Information sharing
• get and set type messages, e.g., HTTP request-reply interaction
• can also be asynchronous, e.g., email messages, chat messages, mobile
phone text messages

b) Task sharing (Task delegation)


• request: description of the task to do
• reply: results performed

11/12/2022 BITS SS ZG531 (Pervasive Computing) 36 of 55

BITS, Pilani - Hyderabad Campus


Interaction Failures

Some of the interaction failures


• Network link failure
• Receiver is down
• Unknown service provider and its location

Way to handle it
• Handle failure with failure status messages

11/12/2022 BITS SS ZG531 (Pervasive Computing) 37 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Intelligent Interaction Protocols

38 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
IS Message Protocols Based
Upon Speech Acts
• Borrowed from Linguistics
• Some speech utterances are like physical actions that
change the state of the world
• E.g., pronouncing someone as ‘man & wife’ in a religious ceremony
• Act: promising, ordering, greeting, warning, inviting and
congratulating.
• Speech Act Theories
• Pragmatic theories of language
• How language is used by people to achieve their goals and intentions

11/12/2022 BITS SS ZG531 (Pervasive Computing) 39 of 55

BITS, Pilani - Hyderabad Campus


Locutionary Model

1) The generation of speech (locution)


2) The choice of speech act (illocution)
3) The intended effect of the speech act on the receiver
(perlocution)

11/12/2022 BITS SS ZG531 (Pervasive Computing) 40 of 55

BITS, Pilani - Hyderabad Campus


Belief, Desire and Intention
(BDI) models
• Belief
• Desire
• Intention

11/12/2022 BITS SS ZG531 (Pervasive Computing) 41 of 55

BITS, Pilani - Hyderabad Campus


Agent Communication
Languages (ACL)
• Standard language for agent communication.
• Proposed by FIPA (Foundation for Intelligent Physical
Agents)
• Also called as FIPA-ACL
• Based on Speech Acts (Communicative Acts)

11/12/2022 BITS SS ZG531 (Pervasive Computing) 42 of 55

BITS, Pilani - Hyderabad Campus


Agent Interaction Protocol
Suite

11/12/2022 BITS SS ZG531 (Pervasive Computing) 43 of 55

BITS, Pilani - Hyderabad Campus


AIPS Example

11/12/2022 BITS SS ZG531 (Pervasive Computing) 44 of 55

BITS, Pilani - Hyderabad Campus


Interaction Protocol Example:
Request

11/12/2022 BITS SS ZG531 (Pervasive Computing) 45 of 55

BITS, Pilani - Hyderabad Campus


BITS Pilani
Hyderabad Campus

Multi-Agent Systems

46 of 55
11/12/2022 BITS SS ZG531 (Pervasive Computing)
Multi-Agent Systems

• Multiple Agents (IS) interacting with each other.

• Properties of MAS
i. Degree of Scale (number of agents)
ii. Type of (Organizational) Control
iii. Homogeneous vs. Heterogeneous
iv. Type of agent interaction (e.g. goal exchange, belief exchange)
v. ACL

11/12/2022 BITS SS ZG531 (Pervasive Computing) 47 of 55

BITS, Pilani - Hyderabad Campus


MAS Application Design

• Agent-Oriented Software Engineering (AOSE)


• Agent-Oriented Development (AOD)
• Proposed by many developers
• Mainly classified into two types:
• Object Oriented Methodologies
• AI Methodologies

• Two model views


• Organisational view
• Operational view

11/12/2022 BITS SS ZG531 (Pervasive Computing) 48 of 55

BITS, Pilani - Hyderabad Campus


Organizational view

11/12/2022 BITS SS ZG531 (Pervasive Computing) 49 of 55

BITS, Pilani - Hyderabad Campus


Operational view

11/12/2022 BITS SS ZG531 (Pervasive Computing) 50 of 55

BITS, Pilani - Hyderabad Campus


Intelligent Agent Interaction

11/12/2022 BITS SS ZG531 (Pervasive Computing) 51 of 55

BITS, Pilani - Hyderabad Campus


Negotiating Resource Access

11/12/2022 BITS SS ZG531 (Pervasive Computing) 52 of 55

BITS, Pilani - Hyderabad Campus


References

• Textbook 1 – Chapter 9

11/12/2022 BITS SS ZG531 (Pervasive Computing) 53 of 55

BITS, Pilani - Hyderabad Campus

You might also like