Upload
bikramjit-singh
View
150
Download
0
Tags:
Embed Size (px)
Citation preview
Repeated Games for Inter-operatorSpectrum SharingBikramjit SinghSupervisor: Prof. Olav TirkkonenInstructor: PhD. Konstantinos Koufos
Department of Communications and Networking (COMNET)Aalto University, School of Electrical Engineering
Friday 24th January, 2014
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
2/18
Outline
I IntroductionI Background
I Related WorksI Related Standards
I Cooperative Spectrum SharingI Repeated Games based Spectrum Sharing
I Proposed Model using Virtual Spectrum PriceI Proposed Model using Mutual Gain/Loss History
I Numerical StudyI Conclusion and Future WorksI References
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
3/18
IntroductionI 1000 times traffic, 50 billion devices by 2020 [1,2]
I 16 billion LTE RAT users in 2018 [3]
I Bandwidth requirements upto 1720 MHz for IMT by 2020 [4]
I
U.S. spectrum surplus/deficit per cell cite [5]
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
4/18
Introduction
I
Summary frequency allocation 0.3-30 GHzI Spectrum utilization variations ranging from 15% to 85% [6]I Utilization of 0.5% in the 3-4 GHz and 0.3% in the 4-5 GHz
bands [7]I Less than 20% utilization in 3 GHz bands [8]I Problem is Spectrum Allocation not Physical Spectrum Scarcity
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
5/18
Background
I Interference managementI How to tackle interference between peer-to-peer entities?
I UtilityI Represents the system’s performance level or QoSI A convex function
I Game theoryI To model interactions involving conflicting objectivesI Game G invloving P number of players, their startegy profile
set S and utility function U , represented as, G = 〈P,S, U〉
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
6/18
Related WorksI Cooperative games
I Highest gainsI Signalling overheadsI Operators trust each otherI Exchange information can be falsified
I Repeated gamesI Operators can be hostileI Punishment mechanismI One-shot games
I Poor EquilibriaI Punishment too strict
I Infinite Repeated gamesI Power domainI Spectrum related work involves monetary transactions
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
7/18
Related Standards
I IEEE 802.11 for Intra-cell/Inter-cell TransmissionI Tackles intra/inter-cell transmission problems
I IEEE 802.11h for Spectrum Management in 5 GHz BandI Only noisy 802.11h device adjusts its spectrum via DFS
I IEEE 802.16h for Improved Coexisting MechanismI Works in the time domain
I IEEE 802.22 for using White Spaces in the TV SpectrumI Intended for centralized inter-network resource sharing
However, Inter-operator spectrum sharing is a non-centralizedimplementation where all the operators participate and aim toresolve inter-operator spectrum needs in the frequency domain
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
8/18
Cooperative Spectrum Sharing
I Motivation - P. Amin, O. Tirkkonen, T. Henttonen, andE. Pernila, "Dynamic frequency selection based on carrierpricing between cells," in Proc. IEEE VTC’13 Spring
I
System architecture
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
9/18
Cooperative Spectrum Sharing
Cooperative decision mechanism for spectrum sharing
I Operator Oi adds unused carrier k and compares its utilitygain Gi,k with operator O−i ’s utility loss L−i,k
if Gi,k > L−i,k thendo START using carrier k
end ifI Operator Oi removes used carrier k and compares its
utility loss Li,k with operator O−i ’s utility gain G−i,kif Li,k < G−i,k then
do STOP using carrier kend if
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
10/18
Repeated Games based Spectrum Sharing
I
System architectureI Game := Game(Utility, Decision, Check)I Spectrum usage favors
I Operator STARTs using the CC alongside opponentI Opponent STOPs using the CC if asked by operator
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
11/18
Proposed Model using Virtual Spectrum Price
I Utility U, chosen as a function of cell throughput T andvirtual carrier price λ,
U = f (T )− λwhere, λ = p1(ep2K/K − 1), with p1 and p2 are pricingconstants, K and K are the numbers representing activeand full CC utilization
I Decision, to a new carrier allocation startegy s adopted byoperator Oi , the following utility conditions must satisfy,
Oi : Ui,s > Ui O−i : U−i,s > U−i
I Operators check the game by limiting spectrum usagefavors h against the maximal limit, surplus S,
Oi : h−i − hi ≤ S O−i : hi − h−i ≤ S
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
12/18
Proposed Model using Mutual Gain/Loss History
I Utility U, chosen as a function of cell throughput T ,U = f (T )
I Decision, to a new carrier allocation startegy s adopted byoperator Oi , operator Oi compares immediate utility gainGi,s against expected loss Li,s, and operator O−i comparesimmediate utility loss L−i,s against expected gain Gi,saccordingly,
Oi : Gi,s > Li O−i : L−i,s < G−i
I Operators check the game by limiting spectrum usagefavors h against the maximal limit, surplus S,
Oi : h−i − hi ≤ S O−i : hi − h−i ≤ S
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
13/18
Numerical Study
101
102
103
0
0.2
0.4
0.6
0.8
1
User Rate (Mbps)
CD
F
Orthogonal, Op. AOrthogonal, Op. BFull Spread, Op. AFull Spread, Op. BCooperative, Op. ACooperative, Op. BProp., Op. A, S=2Prop., Op. B, S=2
Figure: Rate distribution for operator Oa, Na = 25 users and operatorOb, Nb = 5 users using various schemes in a high interferenceenvironment. The maximum number of outstanding favors, S = 2.
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
14/18
Numerical Study
101
102
103
0
0.2
0.4
0.6
0.8
1
User Rate (Mbps)
CD
F
OrthogonalFull SpreadCooperativeProp., S=2
Figure: Rate distribution for operator Oa with temporal load variations(for first half of simulations with Na = 25 users and the second halfwith Na = 5 users) using various schemes in a high interferenceenvironment. The maximum number of outstanding favors, S = 2.
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
15/18
Numerical Study
101
102
103
0
0.2
0.4
0.6
0.8
1
User Rate (Mbps)
CD
F
CooperativeProp., S=2Prop., S=4
Figure: Rate distribution for operator Oa, Na = 25 users using variousschemes in a high interference environment. The maximum numberof outstanding favors is varied and rate curves are analysed for,S = 2 and 4.
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
16/18
ConclusionI Investigated the impact of noncooperative games between the
operators on spectrum sharing
I Properly modelled games provide a clear gain over full spreadand orthogonal spectrum sharing
I Gains are significant when the number of serving users by anoperator is relatively large
Future WorksI Cooperative schemes for intra-operator radio resource
management
I Using outstanding favors in the decision making process
I Load prediction
Repeated Games for Inter-operator Spectrum SharingFriday 24th January, 2014
17/18
References
[1] "Wake-up call: Industry collaboration needed to make beyond 4G networks carry1000 times more traffic by 2020," White Paper, Nokia Siemens Networks, Aug. 2011.[2] "More than 50 billion connected devices," White Paper, Ericsson, Feb. 2011.[3] "LTE release 12 - taking another step toward the networked society," White Paper,Ericsson, Jan. 2013.[4] "Estimated spectrum bandwidth requirements for the future development ofIMT-2000 and IMT-Advanced," International Telecommunication Union ITU, Tech. Rep.M.2078.[5] Federal Communications Commission FCC, Tech. Rep., Feb. 2012.[6] "First report and order in the matter of revision of part 15 of the commission’s rulesregarding ultra wideband transmission systems," Federal CommunicationsCommission FCC, FCC 02-48 ET Docket 98-153, Feb. 2002.[7] D. Cabric, S. M. Mishra, and R. W. Brodersen, "Implementation issues in spectrumsensing for cognitive radios," in Proc. Asilomar Conference on Signals, Systems andComputers (ASILOMAR’04), Nov. 2004, pp. 772-776.[8] V. Valenta, R. Maršálek, G. Baudoin, M. Villegas, M. Suarez, and F. Robert, "Surveyon spectrum utilization in europe: Measurements, analyses and observations," in Proc.IEEE International Conference on Cognitive Radio Oriented Wireless Networks(CROWNCOM’10), Jun. 2010, pp. 1-5.