Context-Aware Computing Network Architectures, NCA/ECA, and a Privacy Protection Technology EMAPP - PowerPoint PPT Presentation

Loading...

PPT – Context-Aware Computing Network Architectures, NCA/ECA, and a Privacy Protection Technology EMAPP PowerPoint presentation | free to download - id: 1e736d-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Context-Aware Computing Network Architectures, NCA/ECA, and a Privacy Protection Technology EMAPP

Description:

nearest subway station, subway fare and walking route on the map of Paris ... Application servers may enjoy higher performance over time, assuming continuous ... – PowerPoint PPT presentation

Number of Views:16
Avg rating:3.0/5.0
Slides: 65
Provided by: Kami158
Learn more at: http://mmlab.snu.ac.kr
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Context-Aware Computing Network Architectures, NCA/ECA, and a Privacy Protection Technology EMAPP


1
Context-Aware Computing Network Architectures,
NCA/ECA, and a Privacy Protection Technology
EMAPP
National Institute of Informatics (NII),
Japan Shigeki Yamada and Eiji Kamioka shigeki,
kamioka_at_nii.ac.jp http//www.nii.ac.jp/index.html
2
National Institute of Informatics (NII) at a
Glance
  • Established in April, 2000 as one of the
    inter-university research institutes by the
    Ministry of Education, Culture, Sports, Science
    and Technology of Japanese Government
  • Became a new corporate body and affiliated with
    Research Organization of Information and Systems
    in April, 2004
  • Broad range of researches in informatics research
    areas from basic theories to applications
  • Foundations of Informatics, Infrastructure
    Systems, Software, Multimedia Information,
    Intelligent Systems, Human and Social
    Information, and Information Research
  • Close cooperation with industries and
    universities to promote utilization of the
    outcomes of research in our society

3
NII at a Glance (Contd.)
  • Providing a scientific information infrastructure
    in Japan
  • SINET (Science Information Network) and
    Super-SINET
  • Inter-university library-related services such as
    catalog information service (NACSIS-CAT) and
    electronic library service (NACSIS-ELS)
  • Started a PhD graduate school in 2002 currently
    50 PhD candidates
  • Annual budget of 100 Million Dollars and 340
    members, including 170 researchers

4
Topics in this Presentation
  • Our Activities of Research on Ubiquitous
    Computing Networks
  • Context-Aware Computing Network Architectures
    NCA/ECA and their Preliminary Network Performance
    Evaluation
  • Privacy-Protection Technology, EMAPP
    (Encapsulated Mobile Agent-based Privacy
    Protection)

5
(1) Context-Aware Computing Network Architectures
and their Preliminary Network Performance
Evaluation
  • Definition
  • Service scenario examples
  • Requirements
  • Functional modules
  • Network architecture
  • Preliminary performance evaluation
  • Future work

6
Ubiquitous Computing
  • Next Generation Mobile Computing Networks will
    evolve into ubiquitous computing networks
  • Concepts of Ubiquitous Computing
  • Ubiquity Everywhere
  • Invisibility Computers will disappear below the
    threshold of our awareness.
  • Invisibility requires computers to capture and
    analyze the user痴 context Context-awareness

7
Context-Aware Application Model
Implicit Input
http//lieber.www.media.mit.edu/people/lieber/Teac
hing/Context/ Out-of-Context-Paper/Out-of-Context.
html
8
Context-Aware Service Scenario 1
Tour guide in Paris
  • You are walking in Paris
  • You ask your wearable computer how to get to the
    Eiffel Tower
  • The context-aware system automatically captures
    your context
  • your voice (lt- microphone)
  • your location and direction (lt- GPS)
  • analyzes and interprets the context
  • retrieves the route information
  • displays the information on your HMD (Head Mount
    Display)
  • nearest subway station, subway fare and walking
    route on the map of Paris

I want to visit the Eiffel Tower!How can I get
there?
9
Context-Aware Service Scenario 2
  • You are sleeping in an airplane
  • Your boss phones you at your office
  • The context-aware system detects the call and
    analyzes your context
  • your location
  • your activity status
  • available communication devices
  • understands
  • you cannot answer the call as you are asleep
  • you have a headset-type device near your seat
  • stores the phone message in a voice mail server
  • sends a message to your handset device
  • You read the message when you wake up

Personal communication service
--- Context --- Location on an airplane Activity
status sleeping Available device headset
10
Previous Researches on Ubiquitous Computing
  • Most of the research efforts for the
    context-aware computing have been devoted to
    sensor devices, computer-human interactions, and
    context-aware application software
  • Few research works focus on the networking issues

11
Our Approach on Ubiquitous Computing Networks
  • To spread context-aware services widely, we need
    a universal context-aware service network
    infrastructure
  • Context-aware services should be available
    anywhere but not restricted in local areas
  • We should make use of the WAN痴 capability of
    covering both the small and wide areas and
    enabling seamless network connection

12
Functional Requirements of Context-Aware Services
  • Context-Awareness
  • Captures,stores, and updates the user context
  • Information Binding
  • interprets the user context and associates it
    with the appropriate contents and services that
    the users wants
  • Information Provision
  • Finds the location of contents and services,
    makes an access to them, and converts them to be
    adapted to the information receiver痴
    communication environment if necessary
  • Provides the contents and services with the
    information receiver at the right timing

13
Context-Aware Service Category and Service Model
Context-Aware Personal Communication Services
Communication Context
Communication Context
Network
Communication Receiver
Communication Sender
Context Analysis, Conversion
Media and Services
Media and Services
Context-Aware Information Delivery Services
User Context
Network
User
Content Server
Context Analysis, conversion
Contents And Services
Contents and Services
14
Consideration on Context-Aware Network
Architecture
  • Context-aware services must seamlessly be
    available
  • indoor --- outdoor
  • LANs --- WANs
  • stationary --- mobile
  • Wide-area coverage and mobility management
    capabilities must be supported
  • Network Architecture that integrates a 3G network
    architecture with a wireless LAN (WLAN)

15
Functional Modules for Context-Aware Services
  • Four functional modules for context-aware
    services
  • to handle the user context and the relevant
    content
  • (1) UIN (Universal Information delivery
    Navigator)
  • decision making module
  • collects user context from user devices
  • executes user authentication
  • analyzes and interprets the context
  • decides the content that the user needs
  • refers to a directory server to obtain the
    location of the content
  • sends the content itself or the content location
    information to the user

16
Functional Modules for Context-Aware Services
(Contd.)
  • (2) DA (Directory Agent)
  • Manages the location of content servers connected
    to the Internet
  • Processes service discovery queries from the UIN
  • Responds to the UIN with service replies
    including the location of the content server
  • (3) UMD (User Management Database)
  • Manages information about users (e.g.,
    authentication information, preferences and user
    context information that users have sent in the
    past)
  • User context information is updated when user痴
    context changes
  • (4) MPS (Media Processing Server)
  • Converts the contents and services provided by a
    content server into an appropriate form, if
    necessary

17
Access Network for Context-Aware Services
  • Integration of UMTS release 5 (all IP network
    architecture) and Wireless LAN

WLAN
ISP Network
MT
AP
R
Internet
R
PSTN/ISDN
PS-CN
UTRAN
HSS
IMS
RNC
BS
MT
SGSN
P-CSCF
GGSN
S-CSCF
I-CSCF
UMTS Release 5
18
Disposition of Functional Modules in the Network
  • Disposition of the functional modules in a
    network plays a key part in the context-aware
    network architecture
  • Content servers should be placed on the end-user
    side in local area networks connected to the
    Internet
  • Two alternatives on the location where the four
    functional modules (UIN, DA, UMD, and MPS) should
    be placed

19
Two Alternative Network Architectures
  • Network-Centric Architecture (NCA)
  • stores, analyze, interprets the user context in
    various functional components managed by the
    network operator and connects relevant functional
    components with SIP.
  • End-user-centric architecture (ECA)
  • stores and interprets the user context in the
    end-user痴 functional components managed by
    end-users or service providers and connects
    relevant functional components with HTTP.

20
NCA (Network Centric Architecture)
User information is managed by the UMD through
GGSN
LAN
Internet
R
Content Server
ISP network
WLAN
AP
MT
R
R
HSS
MPS
(4,9)
UMD
(11)
UIN
DA
5,10
12
IMS
PS-CN
UTRAN
13
(3)
6
(8)
(1)
RNC
BS
MT
SGSN
P-CSCF
GGSN
(2)
16
(7)
S-CSCF
15
I-CSCF
14
21
NCA Information Flow from Capturing User Context
to Displaying the Content
22
ECA (End-User Centric Architecture)
LAN
Content Server
Internet
R
WLAN
ISP network
MT
AP
R
R
ECA
R
DA
UIN
MPS
UMD
IMS
PS-CN
UTRAN
GWTI
S-CSCF
RNC
BS
MT
SGSN
GGSN
P-CSCF
HSS
I-CSCF
23
ECA Information Flow from Capturing User Context
to Displaying the Content
24
Preliminary Performance Evaluation of NCA and ECA
  • Need to evaluate the network architecture from
    various viewpoints and criteria
  • Ease of deployment of new services, network cost,
    network flexibility and scalability
  • As a first step comparison of overall network
    performance of NCA and ECA
  • Purpose
  • Not to obtain precise or absolute network
    performance values
  • But to reveal the general characteristics of the
    two architectures
  • to clarify which design parameters have the most
    significant influence on performance
  • Performance model to measure the response time of
    the network

25
Network Performance Evaluation
  • The response time
  • The interval between the time when a user
    generates a user context and the time when the
    user receives the contents/services location
    information.
  • Mean and 95th percentile response times of the
    two architectures are obtained by queuing theory
    and simulation

26
Performance Models
  • The response time can be broken down into three
    delay elements
  • Processing delay for processing SIP/UDP/IP,
    HTTP/TCP/IP, and SLP/UDP/IP packets
  • IP-network delay includes all of the
    network-layer, data link layer and physical layer
    delays
  • Wireless communication delay generated at
    wireless access sections

27
Performance Simulation Parameters
Parameters Traffic Distribution Values for Calculation and Simulation
Wireless-Related Delay either in MT, UTRAN and PS-CN or in MT,WLAN and ISP Network Normal Distribution 10 msec. as an average 3 msec. as a standard deviation
IP Network Delay in the IMS Normal Distribution 10 and 20 msec. as averages 3 and 4 msec. as standard deviations
Network Delay in the Internet Normal Distribution 50 and 100 msec. as averages 7 and 10 msec. as standard deviations
Processing Time in Application Servers Exponential Distribution 10, and 50 msec. per IP packet
Utilization of Application Servers Constant From 0.1 to 0.9 with the interval of 0.1
IP Packet Arrival Rate Poisson Distribution Arrival rates are determined so as to satisfy the specified utilizations of application servers.
Processing Time in MT Constant 10 msec.
28
Server-processing time of 50-ms and Network
Delays of 10-ms for NCA and 50-ms for ECA
29
Server-processing time of 50-ms and Network
Delays of 20-ms for NCA and 100-ms for ECA
30
Analysis in Low-Performance Application Servers
  • For Small Network Delays
  • ECA has smaller response times than NCA has
  • For Large Network Delays
  • ECA has smaller response times than NCA has
    except for the cases with low application-server
    utilization
  • For low performance application servers, ECA is
    better in the response time than NCA because
    server processing time is a dominant factor and
    NCA requires many traversals of application
    servers

31
Server-processing time of 10-ms and Network
Delays of 10-ms for NCA and 50-ms for ECA
32
Server-processing of 10-ms and Network Delays of
20-ms for NCA and 100-ms for ECA
33
Analysis in High-Performance Application Servers
  • For Small Network Delays
  • For Server utilization under 0.7, NCA has smaller
    response times than ECA
  • For Server utilization over 0.7, ECA has smaller
    response times than NCA
  • For Large Network Delays
  • Similar results are obtained
  • The ECA response times suffers from much larger
    delays while NCA response times are almost
    unchanged as in the cases of small network delays
  • For High-performance application servers,
  • When the application servers are busy, NCA is
    better in the response time than ECA
  • When the application servers are not busy, ECA is
    better than NCA
  • This is because network delay is a dominant
    factor and ECA suffers from large internet delays
    while NCA will have smaller network delays
    maintained by the network operator

34
Summary of Performance Evaluation
  • Response times of NCA and ECA greatly changes
    depending upon given network parameter values
  • Average network delays will increase over time
    because of continuous network growth and
    geographical expansions
  • Application servers may enjoy higher performance
    over time, assuming continuous improvement of
    device technologies
  • This implies that NCA will be slightly
    advantageous in terms of response times because
    the NCA痴 response times are mitigated by fast
    servers

35
Future Work
  • Detail component design
  • Comparisons of the NCA and ECA from other
    technical viewpoints such as network scalability
    and reliability
  • Distributed Allocation of Context-Aware
    Functional Components over the Network
  • Incorporation of Privacy and Security Mechanisms
  • Performance Simulation Considering User Mobility,
    Frequent Context Updates and Heterogeneous
    Network Topology

36
(2) Privacy-Protection in Ubiquitous Computing
Environments EMAPPEncapsulated Mobile
Agent-based Privacy Protection
  • Background
  • Privacy model
  • Features and problems
  • Verification by scenarios
  • Overall Architecture
  • Future work

37
Why is Privacy ProtectionImportant in Ubicomp
Environments?
  • Two major concepts of ubiquitous computing
  • Ubiquity
  • Invisibility
  • Invisibility requires context-awareness that
    captures and interprets user context
  • User context includes privacy穆ensitive personal
    data such as user痴 location, activity status,
    and preferences
  • New privacy protection technologies are required
    for ubiquitous environments
  • Dynamic changes of user痴 computing and
    communication environments

38
Why Do We Need New Security and Privacy
Technologies?
  • Two major concepts of ubiquitous computing
  • Ubiquity
  • distributed and ad-hoc in nature
  • devices are not always administered by the same
    entity
  • every device becomes a potential gateway to leak
    information across network perimeters
  • Invisibility
  • requires context-awareness that captures and
    interprets user context
  • user context includes privacy穆ensitive personal
    data such as user痴 location, activity status,
    and preferences
  • New security and privacy protection technologies
    are required for ubiquitous environments

39
Privacy Invasions in Ubicomp Environments
Bob (Data Collector)
Improper use of Alice痴 personal data
Data Collection
Data Copy
Little control over how her data will be used
Unauthorized use of Alice痴 personal data
Alice (Personal Data Owner)
Carol (Data User)
40
Privacy Issues
  • Where to store personal data?
  • End-User Centric Architecture (ECA)
  • Into stationary servers and devices
  • Into wearable servers and devices
  • Network- Centric Architecture (NCA)
  • Who manages privacy?
  • User, Network Operator, or Service Provider
  • How to protect privacy?
  • Existing TechnologiesP3P and pawS system
  • Our approach EMAPP

41
Context Data Storage Management ECA
  • User context data are stored in user facilities
    (UMD) and managed by users or service providers
  • Users feel easy
  • User has all the responsibility

LAN
(18)
Contents/ Services Server
WLAN
R
Internet
(17)
ISP Network
(19)
(20)
MT
AP
R
R
(6)
(11)
LAN
R
(5)
(12)
UTRAN
DA
PS-CN
(8)
(7)
(1)
(2)
(4)
(3)
(10)
(9)
RNC
BS
MT
SGSN
GGSN
UIN
(16)
(13)
(15)
(14)
UMD
MPS
42
Context Data Storage ManagementNCA (Network
Centric Architecture)
  • User context data are stored inside the 3GPP All
    Network (UMD) managed by Network Operators
  • Secure and uniform management
  • Users may feel uneasy

LAN
(23)
WLAN
Internet
Contents/ Services Server
ISP Network
R
(22)
MT
AP
R
R
(25)
(24)
(21)
IMS
DA
(26)
PS-CN
UTRAN
(11)
(12)
(3)
UMD
UCN
(2)
(4)
(5)
(1)
(13)
RNC
BS
MT
SGSN
GGSN
P-CSCF
MPS
(10)
(7)
(8)
(18)
(20)
(17)
(16)
(19)
(6)
(15)
S-CSCF
(9)
I-CSCF
(14)
43
Design Space for Privacy Protection
Data Collector (Service Provider or web site)
Access
Prevention Avoidance Detection
Data Collection
Second Use
Prevention Avoidance Detection
Prevention Avoidance Detection
Personal Data Owner
Data User
44
Classification of Privacy Protection Technologies
(by X. Jiang (UCB))



RBAC
Anonymization Pseudonymization
Prevention
Location Support
Wearables
P3P
Avoidance
User Interfaces for Feedback, Notification, and
Consent
Privacy Millers
Detection
Collection
Access
Second use
45
pawS a Privacy Protection System (ETH)
  • (3) Privacy Policy Download

Privacy Proxy
Service Proxy
(4) Comparison of Privacy Policy with User
Preferences
(5)Personal Data
(2) Personal Data Service Name
(1) Service Announcement
Privacy Beacon
Privacy Assistant
46
Encapsulated Mobile Agent-based Privacy
Protection EMAPP
  • Most of Existing Privacy Models 徹utgoing Data
About PowerShow.com