Context Awareness in Network Resource Management
Download
Report
Transcript Context Awareness in Network Resource Management
Context Awareness in Network
Resource Management
Χρήστος Αναγνωστόπουλος
Communications Network Laboratory
Pervasive Computing Research Group
Department Informatics and Telecommunications
University of Athens - Greece
[email protected],[email protected]
Θέματα:
Έννοια της Πληροφορίας Πλαισίου (Context)
Πόροι Δικτύου (Network Resources)
Διαχείριση Πόρων Δικτύου βασισμένη στην
Πληροφορία Πλαισίου (Context-Aware
Resource Management)
Θέμα Πρώτο
Context…
Πληροφορία Πλαισίου - Context
Context: “Τhe information that can be used
to characterize the situation of an entity,
where an entity can be a person, place,
physical or computational objects”
[Rakotonirainy 2000]
Οτιδήποτε κρίνεται σημαντικό για να
περιγράψει κάτι, την κατάστασή του, και τον
τρόπο συμπεριφοράς του.
Πληροφορία Πλαισίου - Context
Κατηγοριοποίηση
Primary/Physical Context: Location, Identity, Time,
Activity {where, when, what occurs}
User Context: User requirements, profile, charging and
security policies, Position, QoS {who}
Device Context: H/S Characteristics, Location, Connectivity,
OS/Application running, Manufacturer, Network Operator,
Security policy, Owner {what can be used}
Application Context: Data format, Display, Voice,
Connectivity, State {what can be obtained}
Network Context: Bandwidth, Storage, Protocols, Caches,
Signaling
context information is not complete
Πληροφορία Πλαισίου - Context
Relation between contexts:
The location of a device, which belongs to a user, may be
his/her location
A user’s activity depends on his/her location
My PDA is located in Room 1, your laptop is located in
Room 2, and Rooms 1 and 2 are rooms of the 1st floor,
then, my PDA is next to your laptop
Physical
Application
Activity
Social
In Room 22, the number of people is greater than 5 and
a .ppt application is running and Chris’ location is this
User
room, then, the activity in Room 22 is presentation and
speech notes are downloaded to the PDA of each user
located in this room, during that presentation.
Device
Physical
Θέμα Δεύτερο
Network Resources…as Network Context…
Network Resources
Basic Resources (physical entities/measurable quantities):
Bandwidth: The transferred capacity between nodes of
the network. Also, Timeslots and Frequencies
Power: Overall Signal to Noise ratio (SNR)
Storage:The computing power of the network element
Implicit Resources
Cache: Replicating part of the information residing to a
remote server in the local system or in systems
geographically dispersed inside the network
Protocol: Efficient protocol
implementation/configuration (e.g., TCP window size)
(IP vs. mobile IP)
Signaling: Specific protocols for: connecting wirelessly,
handing over,…[Signaling messages consume part of
useful bandwidth]
signaling
Network Resources
power
Health Safety
High Autonomy
User
perspective
R
Blocking Probability
Dropping Probability
Periods of Interruptions
Stable QoS {Allocated bandwidth, BER, Delay}
Operator
perspective
Capacity=#users served simultaneously
High Utilization
Load Balancing
Θέμα Τρίτο
Context…Awareness
Setting up the Scene
Resource Management Improvement of network performance
Resources are specific and can be reused
Mobility
No fixed point of connection
Dynamical resource allocation within mobile networks
Task:
Find specific characteristic that may assist us in the network
resource management
Velocity
Direction
Location
Pattern…
Location Estimation
Tracking and Positioning
Cell Of Origin (COO): The differentiation of each
cell
Time Of Arrival (TOA): The time window between
sending and receiving signal
Angle Of Arrival (AOA): Measures the angle of the
received signal, at least two directional antennas
Signal Strength Measurement (SSM): By the signal
attenuation equation
Categorization of Positioning Systems
Positioning Systems
Infrastructure
Outdoor
Indoor
Separate
Positioning
Infrastructure
Wi-Fi-enabled
Satellite-based
Terrestrial-based
GPS
EGNOS
GLONASS
GALILEO
Information
GSM
WLAN
(Nibble,
Ekahau)
Positioning Systems
Physical Location
Symbolic Location
Tree-based
Spatial Ontologies
COO
TOA
Relative
AOA
Absolute
GPS
accuracy
Location Prediction
?
Knowledge of the future location enables the
network to avert undesirable situations (e.g.,
dropping a call/unavailability of resources)
Required knowledge:
Velocity and Direction
History of Movements/Handover behavior
Time in the cell of residence
Handoff estimation / predictive-adaptive bandwidth
reservation
e.g.,
1.
2.
3.
Bhattacharya and Das (1999) Theory of Entropy (LeZi)
Maguire and Liu (1996) Regularity-pattern detection (Kalman Filters)
Hadjiefthymiades and Merakos (1999) Learning Automaton
(Reward/Penalty)
Location-Aware Resource Management
Exploitation of the terminal instantaneous position information
toward the management of network resources (Synchronous
Management)
Continuous recording of terminal position + historical
movements patterns + velocity/acceleration/direction implies
proactive management of network resources (Asynchronous
Management)
synchronous
asynchronous
proactive
time
Snapshot
Sort Term Management
recording
time
Snapshot
Long Term Management
Short Term Resource Management (SRM)
Admission Control:
Context = {Users position, Idle/Active sessions, Roaming in the
current cell}
Event = {User on the boundary of two or more cells}
Action= {Refuse the call initiation. This can be handled by an
adjacent base station}
Short Term Resource Management (SRM)
Network Reconfiguration:
Context = {Users position, Idle/Active sessions, Roaming in a cluster of
cells}
Enhanced Context = {Calculation of the load in each cell, e.g., session
initiation/termination probabilities}
Event = {Some cells are congested}
Action= {Re-organization/reconfiguration of resources.
(i) Frequencies are temporarily borrowed by adjacent cells to cater for increased
load
(ii) Inside a cell with low user density, common channels (e.g., RACH-GSM), is
reconfigured to use less timeslots per time unit
(iii) Rearrangement of user population in order to optimally distribute the load
and maximize utilization
The user is provided with specific proposals on how to reach other cells where traffic
load is less and better QoS can be attained
}
Case Study: MITOS
(Alyfantis, Hadjiefthymiades, Merakos 2005)
Short Term Resource Management (SRM)
Handover:
Context = {Users position, Idle/Active sessions, Roaming in the
current cells}
Event = {User on the boundary of the cell, The load in the
adjacent cell is lighter}
Action= {
(i)The terminal is instructed to switch communication i.e., perform a
forced handover to the indicated base station.
(ii) Forced handover could be the support of specific QoS requirements
of the user (user profile, policy,terminal capability).No physical
relocation is required
}
Long Term Resource Management (LRM)
Fine-grained pre-reservation of resources:
Goal={Overcoming HO-induced problems. The user does not experience
service discontinuations or low service quality. Reduction of HO blocking
probability}
Context = {User in the current cell}
Action= {Find the most likely neighbor cell(s) of the current cell and
reserve resources in those cells}
Result={
(i) Too early reservation leads to undesired waste of resources and low network
utilization.
(ii) Delayed pre-reservation may en-up with few resources than required, forcing
the termination of sessions and low experienced QoS
}
No advance
reservation in
candidate cells
[No HO provision]
Less advanced
Advance
reservation in
all cells
[Crude HO provision]
Wireless Resource Management
Advance
reservation in
most likely cells
[Direction Sensitive
HO provision]
More advanced
Long Term Resource Management (LRM)
Protocol Management:
Context = {Different networks (GSM/WLAN), where the user can
have access, User position, Spatial information about the network
resources, like protocol stack/components/modules (4G
infrastructures), Dual protocol stacks is not feasible}
Event = {Switch from one network to other (TCP variant GSM to
plain-vanilla TCP/WLAN)}
Action= {
(i) Advance protocol reconfiguration (downloading) to cater for another
network, Push components, which will be requested by the terminal, to
its nearest nodes/access points/base stations
(ii) Tune protocol parameters w.r.t. current location and network status
}
Case Study: Museum Wearable
(Sparacino 2002)
More context…
Location
User profile {Charging options, Preferences, Personalized services}
Device profile {Capabilities}
Service profile {LBS, software}
…
From Mobile Computing to
Context-Aware Computing to
Ubiquitous and Pervasive Computing
http://p-comp.di.uoa.gr