Integration of WiMAX and WiFi
Optimal Pricing for Bandwidth Sharing
Dusit Niyato and Ekram Hossain, TRLabs and University of Manitoba
IEEE Communications Magazine • May 2007
報告者：李宗穎
Outline
• Introduction
• Major Research Issues and The Related
Approaches
• Pricing for Bandwidth Sharing in An
Integrated WIMAX/WIFI Network
• Conclusions
2
An integrated WiMAX/WiFi network
3
Protocol Adaptation and QoS Support
• 802.16e
– unsolicited granted service
– polling service
– best effort service
• 802.11e
– low-priority traffic
– high-priority traffic
4
QoS support in an integrated
WiMAX/WiFi network
• per-flow approach
– guarantee QoS for individual flows
– complexity is high
• aggregate approach
– reduce this overhead by grouping multiple flows
with similar QoS requirements together and
servicing them as a single traffic class
5
Pricing
• pricing issue relates to the control of radio
resource usage from an economic point of
view
– Optimization-Based Pricing
– Game-Theory-Based Pricing
6
Optimization-Based Pricing
• Goal : maximize utility
– the wired network to maximize system utility
• the rate is a function of price
– price-based distributed algorithm for rate
adaptation in wireless networks
• both rate and reliability performances
• Disadvantage
– may not satisfy all the related entities individually
7
Game-Theory-Based Pricing (1/2)
• game-theoretic formulation aims at providing
individually optimal solutions
– suitable for systems with multiple entities
– service providers want to maximize their profit
– users want to achieve their best QoS performance
8
Game-Theory-Based Pricing (2/2)
• Three major components
– the players
– the strategies of the players
– the payoffs for the players
• Nash equilibrium
– no player can increase his/her payoff by choosing a
different strategy
9
System Description
• the WiMAX BSs and WiFi routers are
operated by different service providers
• the WiMAX service provider charges the WiFi
networks with adjustable pricing
• bandwidth sharing and pricing model uses a
genetic algorithm for learning to choose the
best strategy
10
Revenue and Elastic Demand (1/3)
• WiMAX BS charges different prices to different WiFi
APs/routers depending on the bandwidth demand
from WiFi clients
R
(s)
 i 1 [ai ei D(i , bi( s ) )]
N SS
D(λi,bi(s)) : queuing delay
λi : traffic arrival rate
bi(s) : allocated bandwidth
NSS : total number of SSs
ai : indicates the fixed revenue
ei : decreasing rate of revenue due to the queuing delay
11
Revenue and Elastic Demand (2/3)
• a linear demand function expressed as follows
bj(Pk(r)) = cj – djPk(r))
bj(Pk(r)) : the bandwidth demand of node j
Pk(r) : the price charged at WiFi AP/router k
cj : the fixed bandwidth demand
dj : elasticity of the demand function
• The revenue of the WiFi network k is obtained
(r )
k
R
  j 1 Pk( r )b j ( Pk( r ) )
N k( r )
12
Revenue and Elastic Demand (3/3)
• Finally, the cost is calculated from
(r )
k
C
P
( bs )
k

N k( r )
j 1
b j ( Pk( r ) )  Fk( r )
Pk(bs) : the price charged by the WiMAX BS to the WiFi AP/router k
Nk(r) : the number of WiFi nodes served by router k
Fk(r) : a fixed cost for a WiFi router
13
Stackelberg Game and Profit
Maximization (1/2)
• The players
– The WiMAX BS and WiFi APs/routers
• The strategies
– WiMAX BS : the price Pk(bs) charged to the WiFi APs
– WiFi APs : the required bandwidth
• The payoffs
– WiMAX BS and WiFi APs/routers, the payoffs are the
corresponding profits
14
Stackelberg Game and Profit
Maximization (2/2)
• Given the price charged by the WiMAX BS Pk(bs), the
profit of AP k is
πk(r) = Rk(r) – Ck(r)
• WiMAX BS can adjust the price Pk(bs) charged to
router k to achieve the highest payoff

( bs )
R
(s)
 k 1 R
Nr
(r )
k
15
Genetic algorithm for Stackelberg
game for bandwidth sharing
16
Simulation Parameter
BS Type
Frame duration
Bandwidth
Modulation
TDMA/TDD
5ms
20MHz
QPSK (1/2)
SS number
10
WiFi Router Serve Number 4 + 6
17
Profit function of the WiMAX BS
18
Price and bandwidth sharing at the
equilibrium under different traffic
loads at the subscriber stations.
19
Price and bandwidth sharing at the
equilibrium under different numbers of
WiFi nodes served by WiFi router two
20
Conclusions
• Game theory has been used to analyze and
obtain the optimal pricing for bandwidth
sharing between a WiMAX BS and WiFi
APs/routers
21