Cloud2Bubble:Enhancing Quality of Experience in Mobile Cloud Computing Settings
Pedro Maurício CostaIntelligent Systems and Networks
Mobile Cloud Computing & ServicesLake District, 25 June 2012
IntroductionCloud2Bubble Framework
• Smart Environment• Cloud2Bubble interaction loop
Domain of Application• Application on Urban Public Transport• Scenario
Platform Architecture• Domain Management• Rule Engine• Reasoner
Experiment in Public TransportConclusion
Introduction
Pervasive Computing• miniaturisation of devices• modern communication networks• Internet of Things
Human Computer Interaction• shift in paradigm: from PCs to multiple portal-like devices• increases the opportunities to interact with users• Affective Computing, making systems empathic
Smart Environments• making human interaction with the system a pleasant
experience• enhance Quality of Experience
Introduction
Main Data Streams• user & environment generated• service offering, excessive at times • excessive choice may be detrimental to UX and frustrating
Collection of Data
Service Offering
Cloud2Bubble
Cloudabstract computing infrastructure
*-as-a-Serviceaggregates data streams
generates individual profilesaction generation and selection
Bubbleuser-centric, localcomposed of devices and servicescollects user and environment datadelivers relevant services
2
Cloud2Bubble Framework
Continuous loop of interaction between user and system• leverages existing technology and context• integrates emotional context
Delivery of personalised services• enhancing Quality of Experience• influence user behaviour
Criteria for service offering• environment conditions• user preferences• system goal
Cloud2Bubble: Smart System
Smart System• collects user and environment data• assesses its current state• generates and selects actions• acts on the environment accordingly
World Modelling• high-level events• environment model• user profiling
Decision Making• imprecise and incomplete data• fuzzy reasoning
Decision Making
World Modelling
Hardware
EnvironmentPerception
ActionExecution
Cloud2Bubble: Interaction Loop
EnvironmentState
Aggregation
UserProfile
Disaggregation
user data collection stream
personalised service delivery stream
Domain of Application
Urban Public Transport• need for Intelligent Traveller Information Systems• equipped modern networks (sensors, info)• large user base with idle time• potential to enhance Quality of Experience• benefits for both users and providers
Scenario• Maria: daily commuter• has reported noisy journeys• actual journey noise levels recorded• noise levels decrease journey QoE• journey alternatives ranked accordingly• notification of a better alternative
Decision Making
World Modelling
Hardware
EnvironmentPerception
ActionExecution
Platform Architecture
Domain Manager
Rule Engine
Fuzzy Reasoner
Domain Manager
Manages devices and low-level environment state• simple Event-Condition-Action rules• policy-based behaviour specification• propagation of events to the next layer
Ponder2• software framework for ubiquitous systems• facilitates heterogeneous component management• adapts to users’ activity or environment
Public Transport• integrates existing the information system • manages sensors and other environment devices• integrates Maria’s smartphone
Rule Engine
Support for complex rule definition• adds complex event and temporal reasoning• generates actions to take based on system goals, environment conditions
and user profiles
JBoss Drools• shares the world model on a high-level with Domain Manager• hybrid chaining rule engine, Rete-based• builds a network based on the rules, fast-processing
Public Transport• environment update: selects travellers to notify• generates actions to take
Fuzzy Reasoner
Deals with imprecise and fuzzy data• validates actions to be delivered• improves the decision making process
Fuzzy Inference System• deals with approximate reasoning• inference over imprecise or incomplete date
Public Transport• validates actions generated for each user• adapts user profiles given new user updates
Fuzzification Defuzzification
Fuzzy RulesUser Profile(Fuzzy Set)
EnvironmentState (inputs)
User QoE Evaluation
Cloud2Bubble in Public Transport
Public Transport• world model based on sensors and other environment devices• contributes to build a user profile in relation to environment• allows for the identification of (un)comfortable journeys
Travellers’ interactions• user reports identify (un)comfortable journeys subjectively• is notified accordingly to her preferences
Benefits for both travellers and service providers
Policy-based Behaviour Specification
Policy-based action generation• User profile• Environment conditions• System goals
Idea of approaching system goals and individual behaviour
Policy example: lowering carbon-emissions• high-level goals transformed into low-level behaviour • persuade users to change habits
» reduction of CO2 emissions» tailored user influencing
Experiment in Public Transport
Collection of environment and user data• sensor data from smartphone• feedback from user• relationship between environment and user• integration of emotion into the loop
Experiment in Porto, Portugal• 2 week (beginning of June)• 10 travellers• 135 journeys• 30h of sensed data
Early findings• Found the concept very interesting• Would use in return for this valuable service• Users don’t feel the app intrusive
Conclusion
Cloud2Bubble• generic smart platform• user-centric• systemic• applied to Public Transport
Offer services for enhancing Quality of Experience• focus on interaction with user• emotion dimension
Policy-based computing• bringing together high- and low-level dimensions
Next steps• continue the development of Cloud2Bubble• establish the relationship between emotion and QoE
Thank you
Pedro Maurício [email protected] Systems and Networks