Cognitive Engine Development for IEEE 802.22
Lizdabel Morales
April 16th, 2007
Presentation Outline
IntroductionIEEE 802.22Cognitive RadioCE Development ApproachSimulation and ResultsFuture Work
What is a cognitive radio?
An enhancement on the traditional software radio concept wherein the radio is aware of its environment and its capabilities, is able to independently alter its physical layer behavior, and is capable of following complex adaptation strategies.
Cognitive radio
Cognition cycle
Cognitive Radio is a promising tool for
Access to spectrum finding an open frequency and using itInteroperabilitytalking to legacy radios using a variety of incompatible waveforms
Motivation for using cognition in IEEE 802.22 Systems
Using previous experience to predict:Channel reputationIncumbent detectionOther patterns Protect incumbent users by being aware of the environmentCo-existence and self co-existenceSpectrum utilization improvementsFuture proofing for other CR technologies
It is not known whether a CR network can offer satisfactory performance despite the injection of many new incumbent handling mechanisms [Cordeiro, et. al. 2005]
MPRGs Development of an IEEE 802.22 Cognitive Engine
Objective was to create a Cognitive Engine for IEEE 802.22 systemsPhases I and II completedMain Accomplishments:Development of a solid and generic architecture for the IEEE 802.22 CEDevelopment of a flexible framework that allows for future design, development and testing of more sophisticated modules
WRAN Considered Scenario
System DescriptionSingle WRAN BSCPEs with different application requestsIncumbent users TV only and Part 74 devicesEvents that trigger change in the system:New CPE service request in the WRANIncumbent detected in TV channel
House
Cognitive Engine Architecture
Database
Main Controller
REM
Case and Knowledge Reasoner
Multi-objectiveOptimizer
Utility
Spectrum Manager
Channel Modeler and Predictor
Sensing Module
WRAN Cognitive Engine
Cognitive Engine Modules
Sensing Module provides radio environment sensing results REM provides a snapshot of the radio scenario through timeMain Controller decides which algorithm to useCase and Knowledge Reasoner provides coarse solution, starting point for the Multi-objective OptimizerMulti-objective Optimizer further refines the solution obtained by the CBR
Utility function & Performance metrics
Utility function used in CE should reflect the performance metrics of cognitive WRAN systems, and weight of each performance metrics may vary with radio scenarios:
U1 = QoS satisfaction of each (uplink and downlink) connection for adding new CPE connectionsU2 = Incumbent PU protection (fast adaptation and evacuation) more important for relocating CPEs in case PU reappearsU3 = Spectral efficiency more important for multi-cell or large number of CPEsU4 = Power efficiency and interference temperature reduction more important for mobile UE and large-scale cognitive networks
U = w1*U1 + w2*U2 + w3*U3 + w4*U4
Testing scenarios for WRAN BS CE Performance evaluation
Scenario indexNumber of existing CPEsNumber of CPEs to add to networkNumber of initial active channelsNumber of initial candidate channels12319210528310102841020375104037
REM-CKL vs. GA
CKL runs much faster than GA, especially under complicated situations.
802.22 Specification
Current framework picks up after incumbent user is detected
Questions
CPE 1
CPE 2
CPE 3
CPE 4
CH1
CH1
CH2
CH3
WRAN Base Station
TV Station
(Primary User)
051015202530354045
10
0
10
1
10
2
10
3
Number of CPEs Added
Average Adaptation Time [ms]
GA
CKL
WRAN Cognitive Engine
Main Controller
REM
Multi-objective
Optimizer
Utility
Spectrum
Manager
Channel
Modeler and
Predictor
Sensing
Module
Case and
Knowledge
Reasoner