Management Framework for Cognitive Networks Timo Bräysy , Marcos Katz, Matti Latva-aho and Marja Matinmikko
Management Framework for Cognitive Networks
Timo Bräysy,Marcos Katz, Matti Latva-aho and Marja Matinmikko
Research on cognitive networksg
THE current trend in wireless communications a.k.a.:• Opportunistic communications• Autonomous networking
Research topics include:Research topics include:
• Dynamic Spectrum Management and Sharing• Coexistence of primary and secondary users• Multichannel MAC layer issues
• (Cooperative) Spectrum Sensing methods• Resource awareness methods in general
• Adaptive Cooperation Algorithms• e.g. with game theoretic approaches
• Dynamic Topology ManagementDynamic Topology Management• Mesh structures, hierarchies, routing issues
• Autonomous network optimization methods with constraints from constraints from • topology, power efficiency, routing, cost, frequency usage,…
4 Dec 2008 CWC | Centre For Wireless Communications 2
Cognition cycleg y
Processes for:ll ti h icollecting, sharing
and storing of information about the operation environmentManaging external Managing external constraintsdecision making and learningand learningprocessesAdaptationmechanisms
3.12.2008 CWC | Centre For Wireless Communications 3
”Situation management” frameworkgG. Jacobson, Altusys Corp., 2008
3.12.2008 CWC | Centre For Wireless Communications 4
Cognitive Resource Manager Mähönen et al 2006g g Mähönen et al., 2006
Cross layer optimization of heterogeneous
3.12.2008 CWC | Centre For Wireless Communications 5
radios and networks
NEDAT Kant et al., 2008
NetworkEngineeringDesignAnalyticAnalyticToolset
Currentlyincludes: topology topology, routing/QoS and MAC d t ti adaptation
algorithm toolsApplication/extension to cognitive NCW networks!pp / g
3.12.2008 CWC | Centre For Wireless Communications 6
Cognitive Toolbox Modelg
We propose a toolbox model as a conceptual architecture for cognitive wireless networks.
In our proposal the general decision making entity is termed cognitive manager while the entity is termed cognitive manager while the algorithmic implementations of detailed system optimizing tools are cognitive engines.
3.12.2008 CWC | Centre For Wireless Communications 7
Control architecture framework
3.12.2008 CWC | Centre For Wireless Communications 8
Sample case 1: Flexible Spectrum Usep pAutonomous Multioperator
spectrum management
3.12.2008 CWC | Centre For Wireless Communications 9
Sample case 2: LTE A topology managementLTE-A topology management
IP/Ethernet IP/Ethernet Aggregation andAggregation and
TransportTransport
To Core network
eNBDedicated backhaul
Out-of-band relayC A N O P Y
C A N O P Y
TransportTransport
L
eNBeNBeNB
C A N O P Y
In-band relay
eNB
L
eNBL
eNBeNB
C A N O P Y
C A N O P Y
In-band relay
eNB
eNB
femto e-NB
In band relay
3.12.2008 CWC | Centre For Wireless Communications 10
Figure from ”LTE-A technical proposals” by Motorola, April 2008
Sample case 3: SON in LTE-Ap
List of requirements for LTE-A SONSelf deploymentSelf-deployment
Autonomic HW/SW installationAutonomic parameter configurationR l d t l ti i tiRelays and topology optimization
Self-optimization (in operation)Optimization with adaptive AMC, PC, HARQJoint allocation of resources (space/time/freq./code/power)Cooperative eNBs
Self-maintenanceSelf-healing
3.12.2008 CWC | Centre For Wireless Communications 11
Summary...y
We have presented a hierarchical toolbox model as a management toolbox model as a management framework for cognitive wireless networks.networks.
The framework is needed as a starting The framework is needed as a starting point in designing and developing realistic and implementable algorithms realistic and implementable algorithms for future cognitive networks.
3.12.2008 CWC | Centre For Wireless Communications 12
...but where are the blocks...
Centralized Centralized •E.g. CM and CE are operating in central
titentity.Wireless link
3.12.2008 CWC | Centre For Wireless Communications 13
...or perhapsp p
Fully distributedFully distributed• All building blocks of the cognitive All building blocks of the cognitive
management (sensors, CM, CE) are implemented locally (in each UE)implemented locally (in each UE)
• Conflicts of interests between ”players” arise!players arise!
• Cooperative algorithms are neededTh ff t f lfi h lt i ti • The effects of selfish or altruistic nodes must be addressed!
3.12.2008 CWC | Centre For Wireless Communications 14
Questions?Questions?
Thank You!