Chapter 3: Context-Aware Computing Flashcards
Describe a context aware situation of a photo application. What information are involved?
Face detection: Detects Faces
Clock: Stores time for each picture
GPS: Receives and stores location for each picture
Brightness: Adjust screen to current brightness
Tilt: Change orientation based on camera position
Give some examples for context and their sensor
Location (Outdoor); GPS Location (Indoor); RFID, WiFi Orientation; Compass Temperature; Temperature Sensor Ambient sound; Microphone Identitiy; Email Time; Clock Activity; Accelerometers
What is the motivation for context aware computing ?
It enables smart behavior because it can act and react with respect to the current situation.
- It can adapt the operation to the context without explicit user intervention
- Increases usability & effectiveness by considering the environmental context
What are economic motivations for context aware systems
- Location based ads lead consumers to visit stores or make purchases
- Context aware computing market surpass 125 billion USD
What is context?
Any information that can characterize the situation of an entity.
What is an entity in the context of context?
An entity is a person, place, or object that is relevant to the interaction between a user and an application.
User and Application are also an entity.
What makes a system context-aware?
If the system uses context to provide relevant information and / or services to the user. Relevancy depends on the users task.
What are the different context categories according to the conceptual classification
- Physical Environment Context
- User context
- ICT Context / Virtual Environment Context
Explain Physical environment context
Refers to some physical world dimension (location, time, temperature)
Explain User Context
Interactions constrained by users:
- by identity
- preferences
- task requirements
Explain ICT Context / Virtual Environment Context
Awareness of the services that are available (internally, externally, local, remote, distributed system)
What are the different context categories according to the operational classification
Persistence
- Static Context
- Dynamic Context (Source: Sensed, Derived, Profiled)
- ——Other Literature——-
- Primary Context
- Secondary Context
Explain Static context
Context is created and never or rarely changed e.g. road maps
Explain Dynamic context
Context varies with time. Exhibits different persistence characteristics.
Explain the dynamic context from the source sensed
- Obtained from HW or SW sensors
- Changes frequently
- Often requires transformation of the raw data
E.g. GPS
Explain the dynamic context from the source derived
- Obtained by a derivation function (simple math or AI)
- Precision needs to be considered
E.g. “located-near”
Explain the dynamic context from the source profiled
- Supplied byu ser
- longer-lived / more reliable
E.g. job position
Explain Primary context according to the operational classification
Any information that is retrieved without using existing context. Also no retrieve by sensor data fusion operations.
e.g. GPS sensor readings as location information
Explain Secondary context according to the operational classification
Information that can be computed using primary context. By sensor data fusion or data retrieval operations (web service calls)
e.g. List of friends based on personal identity as primary context
Describe the Architecture of a generic context-aware system
Three layers: Context-Aware Applications (update application state in context model) Context Model (query / notify context-aware applications) Physical World (has sensors that update the context model)
Categorize Context-Aware Applications by level of user interaction
- Personalization (e.g. Smart Home)
- Passive context-awareness (Time zone change, ask user to adjust time to new time zone)
- Active context-awareness (Time zone change, adjust time to new time zone)
Categorize Context-Aware Applications by supported features
- Presentation (of information/services to user based on context)
- Automatic execution (of a service for a user based on context)
- Tagging (of context to information for later retrieval)
Give an example of a context-aware application that supports the feature presentation
Multi-model interfaces (change presentation based on):
- Device properties (screen resolution)
- User Preferences (auto-rotate)
Navigation (dynamic information):
- construction work
- traffic jam
Give an example of a context-aware application that supports the feature automatic execution
Sentient Computing:
- physical environment (light) reacts according to user profiles
Spatial events:
- Notify when near shop
Teleporting:
- user interfaces follow user and use nearest device
Give an example of a context-aware application that supports the feature tagging
Apply virtual objects (information) to physical objects
Augments the environment: E.g.
- Virtual post-its: attached to location
- Tourism: show extra info based on location
- Games: geolocate objects (Pokemon Go)
What are the steps of the Context Management Lifecycle?
- Context Acquisition
- Context Modelling
- Context Reasoning
- Context Dissemination
What sources are available for the context acquisition?
- Application State
- Sensors: Device, module, or system that detects physical / logical phenomena
- Sensor Fusion: Software combines data from several sensor to improve application/system performance
Describe three different sensors.
Physical sensors:
- tangible
- generate raw sensor data (primary context)
- modern devices have usually several of them
Virtual sensors:
- no physical presence
- retrieve data from many sources and publish it as sensor data
e. g. calendar, twitter status
Logical sensors:
- combine physical+virtual sensors
- produce more meaningful information
Give an example for a context-aware app running on a virtual sensor
If this then that:
Twitter sensor: get an email when there is a tweet from a specified person
Instagram sensor: share photos on twitter if post includes a specific #
Give an example for a context-aware app running on a logical sensor
Webservice to provide weather information
- Weather station: physical sensors collect information
- Virtual Sensors: maps, calendars provide additional information
Explain Push and Pull operation in the context acquisition stage
Pull: Software makes request to sensor
Push: The sensor pushes data to the context software component
Explain the different event frequencies in the context acquisition
Instant / Threshold violation:
- Events occur instantly
- Sensor data is acquired when event occurs
e. g. State of Bulb, Motion Sensor
Interval / Periodically:
- Events span a period of time
- Sensor data is acquired periodically
Both push and pull methods can be used for Instant and Interval events
Describe the Context Model
- identifies a concrete subset of the context that is realistically reachable from sensors, applications, and users.
- usually explicitly specified by the application developer
Is composed of context attributes:
- Identifier
- Type
- Value
- (optionally specific properties)
What different context modeling approaches exist
- Key value models
- Object based model (composed of Entities, Attributes, Associations)
Describe the key value modeling approach.
- Key value pair of (environmental variable, value)
e. g. (temperature, 20) - Simplest data structure for modeling context
- Easy to manage
- Low expressiveness (no hierarchies / relationships possible, not scalable)
- Frequently used in distributed service frameworks (List of attributes to describe services, simple matching algorithms)
Explain the context lifecycle stage context reasoning.
- a method to deduce new knowledge based on the available context (refers to step 3)
Steps:
- Context pre-processing (cleanse collected sensor data)
- Sensor data fusion
- Context inference (Generate high-level context information using lower-level context in single / multiple iterations)
How is information categorized at the stage context reasoning
Data Range (Low-level context):
- Information without interpretation
e. g. what the network sees
Context (somewhere in the middle):
- information related to the situation of the user / real world
- needed to determine situation
Situation (high-level context):
- information that can be directly used by context-aware applications for adaption
Name some inference techniques for the stage context reasoning
Most of the inference techniques originated from AI:
- Decision Tree, Naive Bayes, SVM, KNN
Give an inference example for activity recognition.
There is a API for activity recognition.
Uses: accelerometer, barometer, bluetooth and WiFi
Determines: activity of user (walking, running, biking, in vehicle)
Explain context prediction for the stage context reasoning
- Reasoning about the future
- Enables proactive adaption:
- prepare for upcoming context events
- optimize adaption over time
Explain how context can be delivered to applications by using a query
Query:
- context-aware application makes requests
- context management system uses queries to produce results
To use when sensors don’t know when to send data to user or when data is not monitored continuously
Explain how context can be delivered to applications by using a subscription
Subscription (publish / subscribe)
- context-aware application subscribes to context management system with requirements
- system return results periodically or when event occurs
Can be used when sensors know about when to send the data and have enough processing power and knowledge
