MOBS 2013
A Model of Quality-in-Use for
Service-Based Mobile Ecosystem
May 2013
Hyun Jung La and Soo Dong Kim
Department of Computer Science
Soongsil University
Seoul, Korea
{hjla80, sdkim777}@gmail.com
A Model of Quality-in-Use for Service-based Mobile Ecosystem
http://soft.ssu.ac.kr
1
Table of Contents
 Introduction
 Related Works
 Service-based Mobile Ecosystem (SME)
 Architecture of SME
 Self-Stabilizing Management Process
 Metrics for Quality-in-Use in SME
 Task Time & Task Completion
 Managing SME with Quality Metrics
 Experiment Results
 Conclusion
A Model of Quality-in-Use for Service-based Mobile Ecosystem
2
A Model of Quality-in-Use for Service-based Mobile Ecosystem
3
Features of Mobile Devices
 Diverse Networking
Supported
 Wi-Fi, 3G, etc.
 Available Everywhere
 Mobility Supported
 Location-Sensitive
 Supporting Diverse Sensors
 Accelerometer, Light Sensor,
 Limited Computing Power
 Limited Resources
 Main Memory, Second
Memory, Bus Size, etc.
 Constrained I/O Devices
 Small Screen Size
 No Physical Keyboard
 No Mouse Support
Orientation Sensor,
Temperature Sensor, etc.
 Additional I/O Devices
 Speech Recognizer, Touch
Screen, etc.
A Model of Quality-in-Use for Service-based Mobile Ecosystem
4
Research Motivation (1)
 Emergence of Services and Mobile Computing
 A solution to remedying the resource constraints while
utilizing benefits of service-oriented computing
 Issues / Challenges in Service-based Mobile Computing
 Unstable Quality of Services (QoS)
 Difficulties in Managing Quality of the Heterogeneous and
Black-box Services
 Limited Network Bandwidth for Mobile Devices
A Model of Quality-in-Use for Service-based Mobile Ecosystem
5
Research Motivation (2)
 SME as the Solution to the Challenges
 Equipped with autonomous service management process
 To realize SME with the autonomous service management
process,
 Inevitable to monitor the quality of services so as to detect
problematic situations of deteriorating overall quality
 Correspondingly, need to devise a quality model for SME
 In this paper,
 To present a model for measuring quality-in-use for services in SME
 By focusing on two quality attributes of quality-in-use as defined in ISO
9126; Productivity and Effectiveness
 To define a set of quality metrics for measuring the two quality
attributes by considering the need for self-stabilizing SME.
A Model of Quality-in-Use for Service-based Mobile Ecosystem
6
A Model of Quality-in-Use for Service-based Mobile Ecosystem
7
Research Work (1)
 Mei’s Work
 To present an approach to quantitatively measuring end-to-
end quality of transaction-based services
 Response Time and Download Time
 Xiong’s Work
 To present a computer service performance model for the
cloud infrastructure
 To estimate performance of service, in terms of response
time
 Deora’s Work
 To present a quality of service management framework based
on user expectation
 To consider functionality, conformance, and reputation
A Model of Quality-in-Use for Service-based Mobile Ecosystem
8
Research Work (2)
 Limitations
 To only consider external quality of the services
 Not to consider the characteristics of services and service-
based mobile ecosystem
A Model of Quality-in-Use for Service-based Mobile Ecosystem
9
A Model of Quality-in-Use for Service-based Mobile Ecosystem
10
Architecture
SNode2
MNode1
App1
MNode2
App2
S25
S23
SME
Agent
SME
Agent
S24
SME
Agent
SME
Agent
SME
Agent
S11
App3
S91
S61
SNode1
A Model of Quality-in-Use for Service-based Mobile Ecosystem
SME
Manager
S21
S12
SME
Repository
S22
S92
SNode3
11
Self-Stabilizing Management Process

Evaluating
Quality of SME
Acceptable?
[Accepted]
Loading
SME Initial
Configuration
[Not Accepted]
A Model of Quality-in-Use for Service-based Mobile Ecosystem

Defining Quality
Improvement
Plan

Taking Quality
Improvement
Actions

Reconfiguring
SME
12
A Model of Quality-in-Use for Service-based Mobile Ecosystem
13
Quality Model of SME (1)
 Purpose of the Quality Model
 To check current status of SME so that SME manager can
manage a consistent level of SME quality
 Criteria of Defining Quality Model
 Quality should be quantitatively measured at runtime .
 Quality metrics should represent current quality objectively.
 The quality model should reflect different concerns of
different participants.
 Important Quality Attributes considered in SME
 Task Time for Productivity
 Task Completion for Effectiveness
A Model of Quality-in-Use for Service-based Mobile Ecosystem
14
Quality Model of SME (2)
 Three Levels of Quality Attributes
Mobile «invokes»
Application
Service
Instance
«realizes»
Service
Type
«managed in»
SME
Mobile
«affects»
Application
Service «depends on» Service
Instance
Type
Quality at
App or Service Instance
Level
A Model of Quality-in-Use for Service-based Mobile Ecosystem
Quality at
Service Type
Level
«depends on»
Quality at
Entire SME
Level
15
Measuring Task Time (TT) (1)
 TT in ISO/IEC 9126-4
 A quality attribute evaluating how long a system or
functional unit takes to complete a task for the given input
 TT in SME
 To consider both the Network Time for sending requests and
receiving results, and the actual Computation Time of the
service instance
k
TT App (Node x  App i , S j ) 
k
k
NetworkTim e(Node x  App i , S j )  Computatio nTime(S j )
A Model of Quality-in-Use for Service-based Mobile Ecosystem
16
Measuring Task Time (TT) (2)
Meaning
TT at Service
Instance Level
TT at Service
Type Level
TT at SME
Level
To represent how fast the
service instance returns its
results to mobile
applications during the time
period
To represent how fast all the
service instances for a
service type return their
results for the given period
of time
To represent how fast all
service types in current SME
provide their services for the
given period of time
A Model of Quality-in-Use for Service-based Mobile Ecosystem
Metrics
∑ TT APP , S 
n
k
 
TTServiceIns tance S j
k
i
App
i 1
n
n
( )
TTServiceTyp e T j =
∑ TT
k =1
TTSME =
(S )
k
ServiceIns tance
j
n
n
∑ TT
j
ServiceTyp e
(T )
j
j =1
n
17
Measuring Task Completion (TC) (1)
 TC in ISO/IEC 9126-4
 A proportion of tasks to be completed over a period of time
 TC in SME
 To represent a ratio of returning its results successfully for a
given period of time
A Model of Quality-in-Use for Service-based Mobile Ecosystem
18
Measuring Task Completion (TC) (2)
Meaning
TC at Service
Instance Level
To measure a proportion of
service invocations made to
a service instance, Sjk, to
deliver successfully results to
mobile applications
successfully
TC at Service
Type Level
To represent how many
invocations are successfully
handled by all the service
instances for a service type
TC at SME
Level
To represent how many
invocations are successfully
handled by the current SME
A Model of Quality-in-Use for Service-based Mobile Ecosystem
Metrics
number of task completed
   total
number of task received
TC ServiceIns tance S j
k
n
( )
∑ TC
TC ServiceTyp e T j =
TC SME =
(S )
k
j
k =1
n
∑ TC
ServiceIns tan ce
ServiceTyp e
n
(T )
j
j =1
n
19
A Model of Quality-in-Use for Service-based Mobile Ecosystem
20
Usages of Quality Metrics
 To give information of current state to figure out
problematic situations
 To evaluate quality stability
 To assess whether quality stabilizing actions solve
potential problems effectively
A Model of Quality-in-Use for Service-based Mobile Ecosystem
21
How to Evaluate Stability (1)
 Three States
 Normal, Poor, Exceeding
 By using two threshold values, determine the acceptance
of quality.
 Thresholdhigh(x) for the high-bound threshold value of x.
 Thresholdlow(x) for the low-bound threshold value of x.
A Model of Quality-in-Use for Service-based Mobile Ecosystem
22
How to Evaluate Stability (2)
 Three States of Each Metric
Thresholdlow(x)
Exceeding
Thresholdhigh(x)
Normal
Thresholdlow(x)
• QoS is poor.
• Needs remedy actions, s.a.
migration and replication
A Model of Quality-in-Use for Service-based Mobile Ecosystem
TTSME
or
TTServiceType(Si)
• QoS is poor.
• Needs remedy actions, s.a.
migration and replication
• Waste of resources
• Consider ROI.
• Reduce
resource consumption.
Poor
Poor
Thresholdhigh(x)
Normal
Exceeding
• Waste of resources
• Consider ROI.
• Reduce
resource consumption.
TCSME
or
TCServiceType(Si)
23
How to Evaluate Stability (3)
 QoE and QoS Level Determination with TT and TC
TC
High
(i.e. high
throughput)
Quality satisfied,
but resource
consumption in
doubt
Can be considered as an ideal area
after additional examination
Thresholdhigh
Ideal
Thresholdlow
Low
(i.e. low
throughput)
Low
(i.e. faster response)
Faulty State
TT
Thresholdlow
A Model of Quality-in-Use for Service-based Mobile Ecosystem
High
(i.e. slower response)
Thresholdhigh
24
A Model of Quality-in-Use for Service-based Mobile Ecosystem
25
Configuration for Experiment
App1
App2
D
D
App2
App1
S11
App2
D
Desktop Node
S
Server Node
C
Cloud Node
M
Mobile Node
Sij
Service Instance
App1
Global
Manager
Mobile Application
App3
App1
C
App4
D
App3
App3
App4
A Model of Quality-in-Use for Service-based Mobile Ecosystem
2
SD
3
S21
26
Quality of Applications and Services
1
Quality of App
ThresholdHigh
…
…
…
…
0.9
0.8
0.7
App4
App3
App2
App1
0.6
0.5
ThresholdLow
0.4
0.3
0.2
Faulty Case #1
0.1
Faulty Case #2
0
…
1
49
50
51
…
210
211
212
Time (sec)
…
Quality of Service
ThresholdHigh
…
…
…
0.9
0.8
0.7
S12
S21
S11
0.6
ThresholdLow
0.5
0.4
0.3
0.2
0.1
Time (sec)
0
…
49
50
51
…
A Model of Quality-in-Use for Service-based Mobile Ecosystem
210
211
212
…
27
A Model of Quality-in-Use for Service-based Mobile Ecosystem
28
Summary of the Work
 To present quality metrics to assess current states of SME
 To extend 2 quality attributes in ISO/IEC 9126-4
 TT for Productivity
 TC for Effectiveness
 To present a method to evaluate the current stability with
the metrics
 To trigger the further activities of Self-Stabilizing
Management Process
A Model of Quality-in-Use for Service-based Mobile Ecosystem
29
A Model of Quality-in-Use for Service-based Mobile Ecosystem
30