wireless agent-based semantic web interaction heikki helin principal researcher, phd emerging...
TRANSCRIPT
Wireless Agent-basedSemantic Web InteractionHeikki Helin <[email protected]>Principal Researcher, PhD
Emerging Technologies and Innovations, TeliaSonera Finland
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Presentation Outline
• About TeliaSonera
• Motivation
– In which field we are playing…
– What we are trying to achieve…
• Agent-based architecture for nomadic environments
– Agents architecture for such environments
– Communication framework for wireless communication
– Ontologies for nomadic environments
• On-going (or soon-to-be-started) research projects
– National, EU/FP6
• Summary
Heikki Helin: Wireless Agent-based Semantic Web Interaction
The Nordic and Baltic leader in telecommunications
• The merger between Telia and Sonera became a reality on December 9, 2002
• Listed on the Stockholm Exchange, the Helsinki Exchange and Nasdaq Stock Market in the USA
• Net sales (pro forma) in 2003 amounted to SEK 81,772 million (EUR 9,010 million)
• The number of employees was 26,000 on December 31, 2003
– R&D personnel : ~1000
TeliaSonera
Heikki Helin: Wireless Agent-based Semantic Web Interaction
The markets of TeliaSonera International
#3
#1
#1
#1
#1#2
#1#1
#1
#1
#1#4
#2
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Motivation
• New wireless Internet services for our mobile (and fixed) customers
– Not only ”new services”, but also enhance ”old ones” to give better user experience
• And do that
– faster,
– easier,
– cheaper, ...
=> To attract new customers, but more importantly: keep the old customers (and at the same time do some academic research)
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Background
• Multi-agent system implementing a middleware providing (agent-based) applications with tools for adaptation in a nomadic environment
• Nomadic environments enable new ways to access services
anywhere, at any time, and using any device
• Challenges that need to be addressed:
– Varying QoS of the wireless networks
– Limitations of mobile devices
– Contextual variability (location, time, user preferences, ...)
• Adaptation to environment is the key issue
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Mobile/Wireless Environment
• Typical characteristics– Low throughput, long delays, unreliable, ...
– Variability– Disconnected mode of operation is the most common state– Different kind of (wireless) networks
• Seamless roaming will be important in the future
• Actors (agents) in nomadic environments should be aware of these issues– Context-awareness (not only location)
• Currently we are mainly considering long thin networks– GSM, HSCSD, GPRS, UMTS, ...– But not excluding WLANs, BT, LAN, xDSL, ...
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Agents in Nomadic Environments
• Suitable for complex environments
– Internet is a complex environment...
– Wireless Internet is even more complex...
– Invisible Internet...
Ambient networks
Intelligent P2P Pervasive Computing
Semantic Web
Ad hoc networks
Ubiquitous Computing
Distributed Artificial Intelligence
Peer-to-peer
Ontologies
QoS
Heikki Helin: Wireless Agent-based Semantic Web Interaction
FIPA Agent Platform
DF = Directory FacilitatorAMS = Agent Management System
ACC = Agent Communication Channel
Heikki Helin: Wireless Agent-based Semantic Web Interaction
(FIPA) Nomadic Application Support
Knowlegde Sharing
Knowlegde Sharing
Controlling & monitoring Controlling & monitoring
Heikki Helin: Wireless Agent-based Semantic Web Interaction
(FIPA) Nomadic Application Support
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Agent Communication Wireless World
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Layered model of agent communication
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Message Transport
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Message Transport
• How messages are transferred between agents
• Desiderata
– Reliability
– Efficiency
– Dynamic adaptation
• Issues
– Performance problems
– Terminal mobility
– Thin clients
Heikki Helin: Wireless Agent-based Semantic Web Interaction
MTP Results – 1+1; 9600bps (300ms)
0
10000
20000
30000
40000
50000
60000
0,5kb 2kb 5kb 10kb
ms
WAP
MAMAv2
IIOP
HTTP
RMI
Heikki Helin: Wireless Agent-based Semantic Web Interaction
MTP Results – 1+1; 115200bps (300ms)
0
2000
4000
6000
8000
10000
12000
0,5kb 2kb 5kb 10kb
ms
WAP
MAMAv2
IIOP
HTTP
RMI
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Message Envelope
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Message Envelope
• FIPA ”specific” layer
• Defines how messages are delivered
– independent of message transport protocols
• Encoding options
– XML, Bit-efficient, IIOP/IDL, Binary-XML, Java Objects, …
• Bit-efficient encoding similar to bit-efficient ACL
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Message Envelope
0
500
1000
1500
2000
2500
3000
Minimal Envelope Typical Envelope
Byt
es
Bit-Efficient
IIOP
bXML
bXML(Plain)
XML
J ava Object
Heikki Helin: Wireless Agent-based Semantic Web Interaction
ACL Layer
Heikki Helin: Wireless Agent-based Semantic Web Interaction
ACL Encoding
• Options– String (s-expression), XML, Bit-efficient
– Binary-XML (w/ special encoding tokens), Java Objects
• Bit-Efficient encoding is the most efficient– Space-efficient
– ...by definition
– Time-efficient
– much faster
– nice for every application; not only for wireless
– necessity for
– high performance applications
– highly utilized servers
– simple parser appropriate for thin clients
Heikki Helin: Wireless Agent-based Semantic Web Interaction
ACL Encoding (request conversation)
0
500
1000
1500
2000
2500
3000
Send Receive
Byt
es
Bit-efficient
bXML
bXML(Plain)
String
String (deflate)
XML
Object
Heikki Helin: Wireless Agent-based Semantic Web Interaction
ACL Encoding
0
2000
4000
6000
8000
10000
12000
14000
16000
One message 10 Messages
Byt
es
Bit-efficient
bXML
bXML(Plain)
String
String (deflate)
XML
Object
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Content Language Layer
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Content Languages
• FIPA-SL
– All-purpose content language
– S-expression (w/ deflate), XML, Binary-XML (w/ special tokens)
• FIPA-CCL
– Language for constraint satisfaction problems
– XML & Binary-XML
• Results similar to those of message envelope and ACL
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Conversation Layer
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Conversation Layer
• Optimizing/modifying existing conversation protocols?
• Developing new conversation protocols?
• Selecting conversation protocol based on current environment
– low bandwidth simple protocol
not so good end result
– more bandwidth more complicated protocol
better end result
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Ontologies in Wireless World
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Ontologies
• An ontology is an explicit description of a domain:– Concepts
– “Wireless network”, “GSM”, “GPRS”, …
– Properties and attributes of concepts– Each “Network” will have “Operator”, “Location”, “Properties”, …– Every “IEEE802.11a” is a “WLAN”
– Constraints on properties and attributes– The name of an Operator is a string– GSM network code consist of CountryCode and NetworkID
– Individuals (often, but not always)– The GSM network operated by TeliaSonera in Finland
• An ontology defines – a common vocabulary– a shared understanding
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Can’t we just use XML?
–
– –
– – – – – – – – –
For machines, a typical web page looks like this:
Heikki Helin: Wireless Agent-based Semantic Web Interaction
XML Helps...
<name>
–
</name><address>
– –
</address><cv>
– – – – – – – – –
</cv>
XML allows “meaningful tags” to be added to parts of the text
Heikki Helin: Wireless Agent-based Semantic Web Interaction
XML ≠ Machine Accessible Meaning
<>
–
<><>
– –
<><>
– – – – – – – – –
<>
To machines, the tags looks like this…
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Web Ontology Languages
• Ontologies have been active research area for a long time
• Web ontology languages
– Provide ”web-friendly” representation language
– Typically based on RDF Schema
• OWL is currently W3C Recommendation
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Web Ontology Languages (cont.)
• XML is not enough– Ontological modeling primitives missing
• RDF/RDFS is better, but– Only a weak semantic interpretation
– No inference model
• OWL (Web Ontology Language)– Extends RDFS
– Additional vocabulary along with a formal semantics
– More vocabulary for describing properties and classes:– Relations between classes (e.g. disjointness)
– Cardinality (e.g. "exactly one")
– Equality
– Richer typing of properties
– Characteristics of properties (e.g. symmetry)
– Enumerated classes
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Need for Ontologies in Wireless World
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Wireless Network Ontology - Motivation
• Machine-readable (& understandable) descriptions of networks– Currently available only for humans
• 3rd party service providers– ”Available networks at Gent?”
– ”Available networks at this Hotel?”
• Future applications– Context-aware
– ”Available networks near here?”
– “Best available network?”
– ”Available networks near my friend?”
– Any application requiring information about (wireless) networks or their properties
– Almost all wireless world applications?
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Wireless Network Ontology – Overview
• Core Concepts– Network -> Wireless Network -> GSM Network -> ...
– Operator
– (Location)
– ...
• QoS Concepts– Throughput, delay, rtt, …
• Services– GSM Services
– Supplementary services, Tele services, …
– WLAN Services
– ...
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Wireless Network Ontology – Core Concepts
Network
Class
WirelessNetwork
Class
WirelineNetwork
ClassdisjointWith
subClassOf
subClassOf
GSM
Class
subClassOf
GPRS
Class
subClassOf
WLAN
Class
subClassOf
IEEE802.11a
Class
IEEE802.11b
Class
subClassOf
subClassOf
UMTS
Class
subClassOf
QoSProperty
Class
Service
Class
hasService
ObjectProperty
range
hasIdentifier
ObjectProperty
Operator
Class
operatedBy
ObjectProperty
domain
rangeavailableAt
ObjectProperty
domain
Location
Classrange
qosProperties
ObjectPropertydomain domain
Identifier
Classrange
domain
PacketSwitched
Class
CircuitSwitched
Class
subClassOf
subClassOf
subClassOf
subClassOf
WAN
Class
LAN
Class
PAN
Class
BlueTooth
Class
subClassOf
subClassOf
subClassOf
subClassOf
subClassOf
subClassOf
hasGPRS
ObjectProperty
range
subClassOf
Restriction subClassOf
1
onProperty
maxCardinality
range
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Wireless Network Ontology – Services
GSMService
Class
SupplementaryService
Class
TeleService
ClasssubClassOf
subClassOf
CallForwarding
Class
subClassOf
SMS
Class
moSMS
DataTypeProperty
xsd:boolean
range
domain
mtSMS
DataTypeProperty
domain
xsd:boolean
range
Service
ClasssubClassOf
subClassOf
cellBroadcast
DataTypeProperty
domain
xsd:boolean
range
FAX
Class
subClassOf
moFAX
DataTypeProperty
mtFAX
DataTypeProperty
domain domain
xsd:boolean xsd:boolean
range range
...
Class
subClassOf
...
Class
subClassOf
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Wireless Network Ontology – QoS ConceptsQoSProperty
Class
LineRate
Class
subClassOf
Delay
Class
subClassOf
Throughput
Class
subClassOf
hasRateValue
ObjectProperty
RTT
ClasssubClassOf
MeanUpTime
Class
subClassOf
ConnSetupDelay
ClasssubClassOf
TimeQoSProperty
Class
subClassOfhasTimeValue
ObjectProperty
domain
RateQoSProperty
Class
subClassOfdomain
ProbQoSProperty
Class
subClassOf
hasProbValue
ObjectProperty
domain
BER
Class
subClassOf
OmissionRate
Class
subClassOf
FrameErrorRate
Class
subClassOf
ConnSetupFailureProb
Class
subClassOf
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Related Projects @ TeliaSonera
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Related projects @ TeliaSonera
• EU/FP6 projects
– Ambient Networks (IP), Magnet (IP)
– CASCOM (Strep)
• National projects
– CAPNET
– Fuego Core
– Intelligent Web Services
Heikki Helin: Wireless Agent-based Semantic Web Interaction
CASCOM – Overview
Context-Aware Business Application Service Co-ordination In
Mobile Computing Environments
• FP6/2 Strep project
– 414 PM/3 years
– 8 partners
• Project goal:
– Develop, implement, validate, and trial of
– Agent-based service coordination infrastructure for
– Innovative semantic Web service discovery, composition, execution
– Across mobile and fixed Peer-To-Peer service networks
Heikki Helin: Wireless Agent-based Semantic Web Interaction
CASCOM – Main S&T Innovations
• Generic Service Agent Architecture and Prototype Demonstrator
– CASCOM Service Coordination Means
– Declarative Service Description
– Flexible Service Matching
– Dynamic Service Composition
– Secure Service Execution & Monitoring
– CASCOM Intelligent Service Agents
– Service Agent Platform for Nomadic Computing and Ad-Hoc P2P Communication
– Context-Aware Service Provisioning
– Dynamic Service Agent Coalitions
Heikki Helin: Wireless Agent-based Semantic Web Interaction
CASCOM – Use Case Scenario
Ad-Hoc Emergency Medical Care and Ad-Hoc Emergency Medical Care and TransportTransport• Medical information • Patient transport• Insurance coverage
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Intelligent Web Services (IWebS)
• Semantic yellow page services
– How the user finds correct service(s)?
– How the service provider advertises the service it provides?
• Dynamic content
• Service composition
• Support for different terminals
– Desktop, mobile, ...
• http://www.cs.helsinki.fi/group/iwebs/
Heikki Helin: Wireless Agent-based Semantic Web Interaction
IWebS – Sample Scenario
Heikki Helin: Wireless Agent-based Semantic Web Interaction
Summary
• Nomadic environments introduce challenges, such as varying QoS, variations and limitations of the mobile devices, as compared to desktop environments
– Device independence and adaptation are key issues when designing and implementing services for such environments
• Agent architecture with efficient communication
• Ontologies– (Wireless) networks– QoS– Other domains need conceptualization as well
– Services– Users
– Preferences, etc.– Devices
Heikki Helin: Wireless Agent-based Semantic Web Interaction
More information
• Heikki Helin and Mikko Laukkanen: Performance Analysis of Software Agent Communication in Slow Wireless Networks. In Proceedings of the 11th International Conference on Computer Communications and Networks (ICCCN'02), October 2002
• Heikki Helin and Mikko Laukkanen: Wireless Network Ontology. In WWRF#9, July 2003
• Heikki Helin: Supporting Nomadic Agent-based Applications in FIPA Agent Architecture. PhD Thesis. University of Helsinki, Finland. February 2003.