Ubiquitous Computing based on VPW research summary

1
Ubiquitous Computingbased on VPW (research
summary)

• Kyung-Lang Park
• (2006. 7. 26)

2
Contents

• Introduction
• VPW Model
• System architecture
• VPW-based service model
• Evaluation
• Conclusion and future work

3
Introduction

• Ubiquitous computing
• Users can use services anytime, anywhere with
  multiple heterogeneous devices
• Global computing infrastructure support
• Sharable devices, sensors, and objects
• We should personalize them to use services
• Related work
• Location-based approach (GAIA, AURA)
• Local server manage devices in an administrative
  domain
• Not user-centric at all
• Hard to consider users own devices
• Hard to manage devices spared over multiple
  administrative domains
• Ranged approach (Nakajima, Khatib, M-GAIA)
• Manage personal space, a list of devices
  surrounding users
• It is hard to assert those devices are always
  adequate for users
• All are occupied by other users
• Quality is low
• The user is just passing through
• It is Personalized location-based approach
• VPW-based approach

4
Ubiquitous computing
A local server can manage them and resolve
conflicts between users,
5
Ubiquitous computing
A local server cannot manage them. How about 4
local servers?
6
Ranged-based approach More like personal spaces,
but not enough
7
VPW Model

• Virtual Objects (VO)
• Uniform representation of all objects
  participating ubiquitous computing
• Services
• Operations to be delivered to users
• Neighbors
• Other people related to the user

8
State of elements
9
Representation of VPW
XML VPW Profile
Diagram
Equation
10
Spatial model of VPW

• VPW span represents the physical range of VPW
• Users scope of action in probability
• Users can set a VPW span
• Spot, radius, and a probability (pspan)
• Users can set multiple spans
• Probability of each span should be recalculated
• Service areas can be VPW spans
• Union of all the operational ranges of VOs that
  are being used by services
• VPW Spans are used in several parts in managing
  VPW

11
Examples of VPW Spans
Basic VPW span (pspan p0)
Multiple VPW Spans
Multiple VPW Spans (Overlapped)(p0 p0 p1)
Two spans are overlapped(p2 p0 p1)
Service area is a VPW span
12
System architecture
Omitting several components
13
Old version
Virtual Personal World
Personal Terminal
Service Provider
Service Hosting Env.
Service Controller
Sensor
Service Core
Service Activator
S
Gateway Interface
O

Service Negotiation
Personal Agent
S
Personal World Manager
Coordinate
Coordinate
Personal Engines
Adhoc Connection
Object
Device
Managing Shared Objects
Object
Device
Personal World Private World
Shared World
Shared World
Gateway
Shared World

Context Engine (Managing Sensors)
Object Engine (Managing Objects)
Actuator Engine (Managing devices)
Private World
WorldWide Gateway Network Infrastructure
14
Differences

• Focus on POM, especially VPW Manager
• Do not consider local gateway
• POM directly connect to the VODs
• Do not consider context engine
• Simplify application services
• Sensors, objects, and devices are unified into
  virtual objects

15
Virtual object daemon

• VOD is a software that interfaces to a real
  object
• It consists of a device driver and an operation
  module
• Operation module performs pre-defined operations
  and exchange messages with external components
• Communication object provide basic communication
  functions and discovery facilities SLP
• Device profile stores capabilities and
  characteristics of the object CC/PP

16
VPW container

• VPW container is a class instance which has three
  lists of elements which are VOs, services, and
  neighbors.
• When managers decide to add an element into VPW,
  it creates an instance of the element
• State transition from recognized to available or
  intend
• VPW can be accesses only through the VPW
  container
• It stores VPW profiles into the personal database
  periodically

17
VPW Control Interface

• VPW control interface provides a graphical user
  interface to control VPW
• Lists up all the element in the VPW
• Draw spatial models
• VPW spans, service areas, operation ranges
• Users can trigger a service or control virtual
  objects by using VCI

18
VPW Control Interface
19
VO manager

• VO manager discovers VOs, decides whether to
  include them in the container, and changes states
  of VOs
• It only includes usable Vos by using a
  probability function denoted as Puse

• Event GI is that the user gets in the operational
  range of the VO
• Event IS is that the user invokes a service that
  can use the VO
• Event UV is that the service uses the VO

20
Puse

• The event GI and IS are independent
• According to the multiplication rule
• There can be more than on service that can use
  the VO
• PGI is obtained by VPW spans
• PIS is obtained from VPW
• If a service i is intend, PISi can be 1
• P(UVIS) is the probability that the service
  actually uses the VO when it is executed

21
? (minimum puse )

• VO manager calculates puse of all the Vos in the
  recognized list.
• The minimum puse denoted as p is specified in the
  user policy
• If a puse of a VO is higher than ?, it could be
  included in the VPW
• User policies also specify the maximum number of
  VOs in a category and the maximum number of VOs
  in total

22
Service manager

• Service manager manages services
• It discovers services by initializing, looking up
  local service registries, and communicating with
  other service managers,
• When it discovers a service, it retrieves the
  service description and adds it into the list of
  recognized services
• If a user wants to use the service, it could be
  included in the VPW and changed into intend
• Explicit by using VPW control interface
• Implicit by specifying conditions in policies

23
Negotiation phase

• If a service intended, it goes into negotiation
  phase
• Checks the possibility to run the service by
  analyzing the service description and VPW

Example of Service Description
24
Neighbor manager

• Neighbor manager communicate with other neighbor
  managers
• If it discovers a neighbor, it adds the neighbor
  into the list
• If it receive the message for sharing or for
  being exclusive to a service, it changes the
  state to the join

25
VPW-based Service Model

• When a service is triggered, it invoke the
  executable specified in the service description
• Service firstly performs initialization
• retrieve VPWProxy from the VPW container
• Obtains information of VOs assigned for the
  service
• Sends commands to VOs and receives results

26
Example of services
27
Service reconfiguration

• When the VPW changed, VPW container sends signal
  to the application
• Application performs the callback function
• Retrieve update of VPW
• Service could be blocked when the VPW do not
  satisfy the conditions to run the service

28
Sequence diagram
29
Providing services on multiple users

• Service could be provided on multiple users
• It generate a temporal VPW by using set
  operations
• Service run on the temporal VPW

Exclusive mode
Sharing mode
30
Evaluation

• Basic performance
• Initialization time and adaptation time according
  to the size of VPW

31
Evaluation

• Operation times according to number of elements

32
Evaluation

• Simulation-based experiments
• Generates Vos
• Generates services
• Define users
• User policies
• Behavior model
• Generates action lists
• Run action lists
• Monitors annoyance
• Calculate average user satisfaction rate

33
Evaluation

• Parameters

34
Evaluation

• Metrics
• Average user satisfaction rate (AURS)
• Obtain by measuring annoyance (annoyance points)
• Service is delayed or not executed (delayed time
  unites)
• Quality of the service is lower than the users
  request ( of unsatisfied functions / total)
• Conflicts occur (0.1)
• AUSR can be defined as Eq. (9)

35
Evaluation

• VO inclusion rate according to the number of
  neighbors and ?

IR is in reverse proportion of ? of neighbor
affects IR.
36
Evaluation

• Annoyance rate and AUSR according to IR

No neighbor
Five neighbors
37
Evaluation

• Comparison of three approaches

No neighbor
Five neighbors
38
Conclusion

• VPW provide more abundant and accurate
  information of users, so that it helps
  application services adapt their operations for
  the user
• Experimental results show that the proposed
  system has reasonable performance in running
  application services and managing personal spaces

39
Future work

• The paper focused on VO management
• Its very small part of our research!!
• Implementation
• POM
• VOD
• Application services
• Address ambiguous things
• How to discover Vos, services, neighbors?
• Automatic execution
• VPW Profiling
• VPW-based service model
• VO hierarchy
• Conflict resolution

40
Personal schedule (2 weeks)

• Submit the paper to wcm journal (ESCI)
• Implement POM more
• Support
• Jookyoungs paper
• MPI project
• RFID/WSN project
• Study VPW-related research
• Service discovery
• Profiling (Some AI techniques)
• WLAN
• Ontology