Professional Documents
Culture Documents
net/publication/220474680
CITATIONS READS
8 558
2 authors, including:
Martin BH Weiss
University of Pittsburgh
143 PUBLICATIONS 1,103 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Martin BH Weiss on 31 May 2016.
Seungjae Shin*
Division of Business,
Mississippi State University – Meridian,
1000 HWY 19N, Meridian, MS 39307, USA
E-mail: sshin@meridian.msstate.edu
*Corresponding author
Reference to this paper should be made as follows: Shin, S. and Weiss, M.B.H.
(2010) ‘Analysis of mobile broadband competition: 3G vs. WiFi’, Int. J.
Mobile Communications, Vol. 8, No. 5, pp.586–601.
He received his PhD from Carnegie Mellon University. His research interests
are broadly focused on the interaction between technology, public policy and
industry as well as cooperation among competing firms.
1 Introduction
3G WiFi
Technology CDMA 2000/W-CDMA IEEE802.11b/g
Throughput 400∼700 Kbps up to 2 Mbps 4.4∼6.6 Mbps up to 11 Mbps
20∼22 Mbps up to 54 Mbps
Pricing Unlimited pricing Unlimited, per-day, per-hour pricing
Spectrum Licensed band Unlicensed band (2.4 GHz)
Coverage Wide-area Local-Area
Providers Verizon wireless, sprint PCS, AT&T mobile T-Mobile, AT&T, Boingo
In the 3G mobile broadband market, there are two competing technologies: EV-DO
(Evolution-Data Only), as an evolution of CDMA 2000, and High Speed Downlink
Packet Access (HSDPA), as an evolution of W-CDMA. EV-DO was deployed ahead
of HSDPA. In 2004, Verizon Wireless and Sprint PCS positioned themselves as pioneers
in the 3G (EV-DO) mobile broadband market. Although the current maximum
throughput is 2 Mbps, its actual speed is known as 500 Kbps. In 2006, AT&T Mobile,
formerly Cingular, started its 3G service (HSDPA). As of 2006, in terms of number of
cities covered by 3G service, more cities were covered by Verizon Wireless and Sprint
than by AT&T Mobile (Mossberg, 2006). According to mobileroar.com, as of July 2008,
comparing the three major 3G operators, Verizon wireless is the number one 3G
US operator in terms of coverage area. T-Mobile built its 3G wireless network in 2008.
From the technology adoption point of view, CDMA2000-adopted carriers like Verizon
and Sprint are leaders and W-CDMA-adopted carriers like AT&T Mobile and T-Mobile
are followers. In this paper, the authors only focus on EV-DO technology as representing
3G service.
In the WiFi market, IEEE802.11b has been a dominant standard for the last several
years. The WiFi market is currently migrating from IEEE 802.11b to IEEE 80211g, to
provide a higher data bit rate standard (Wexler, 2008). In this paper the authors consider
mixed environments of 11b and 11g as representation of WiFi service. Because the two
Analysis of mobile broadband competition 589
standards are compatible, this enables WiFi networks to continue supporting 11b when
migrating to 11g.
Because WiFi uses an unlicensed spectrum and its service area can be added
spot-by-spot, the market entry barrier is not as high as with 3G. Therefore, it is possible
for WiFi to have a community-based free service or revenue sharing between
infrastructure providers and location owners. However, this paper focuses on the
commercial type of WiFi provider, i.e., Wireless Internet Service Provider (WISP).
As we mentioned previously, T-Mobile, the fourth largest US wireless carrier, has
deployed WiFi networks since 2002. According to T-Mobile’s website,9 it has 10,000
hot-spots in the USA as of 2008. Wayport was one of the leading WiFi based Internet
Service Providers (ISP). After AT&T acquired Wayport in 2008, AT&T began operating
20,000 hot-spots10 in the USA, including McDonald’s restaurants and Starbucks coffee
shops. AT&T is the number one WiFi internet access provider in terms of number
of hot-spots.
As the number of hot-spots increase and 3G coverage area expands, competition
between the two technologies increases in the common target service area of 3G
and WiFi. The authors cannot say that one is better than the other. For those in need
of a faster internet connection, WiFi is a better solution; however, 3G is a better option
for those desiring broader availability and better mobility. Due to the lack of roaming
capability between 3G and WiFi networks, a mobile user is forced to choose one of the
services. Currently, the price for 3G services (Verizon Wireless) is $60 with two-year
subscription agreement and voice subscription. The WiFi services (T-Mobile) require a
one-year subscription agreement with a monthly subscription fee of $30.
4 Competition model
10% to 100%. The αWF is a percentage of WiFi coverage, which is also increased in 10%
increments from 10% to 100%. The αWF value determines the number of WiFi hot-spot,
which determines the number of access point in the WiFi network. These two parameters
make 100 cases (10 p-ratio × 10 αWF), where equilibrium analysis will be made.
Table 3 3G cell radius, estimated population covered per cell site and number of estimated
hot-spots
Number of
3G cell Number of Total potential users maximum
Density radius 3G cell Population/cell (100% penetration) hot-spots
Dense urban 0.5 mile 100 25,000 2.5 million 10,000
Rural/small 5 mile 1 20,000 20,000 100
town
Analysis of mobile broadband competition 591
Henkel et al. (2002) conducted a survey for WiFi based WLAN user preference.
One of their questions was “How interested would you be in WiFi service at the price
indicated below if the monthly fee was (a) $20, (b) $40, (c) $60, and (d) $80?” A similar
survey for 3G service was accomplished by Strategic Policy Research (2001). Table 5
contain survey data of the willingness-to-pay for WiFi and 3G internet access services.
The authors borrowed these distributions from the above two market analysis papers to
estimate demand functions of WiFi and 3G. To make an estimate of demand function, the
authors use a random-number generating simulation because linear and curve estimation
does not provide a good R2 value. The random numbers are generated from the above
empirical distribution of the willingness-to-pay data. The authors generated two random
numbers: one for willingness-to-pay of 3G (W3G) and the other for willingness-to-pay for
WiFi (WWF).
592 S. Shin and M.B.H. Weiss
4.5 Cost
As the authors mentioned in Section 2, 3G network deployment and service provisioning
are centralised planning and operation. In contrast to 3G, WiFi service can emerge in a
decentralised fashion (Lehr and McKnight, 2003). The different deployment strategies
impact the cost structure of both services. The authors assume that 3G service providers
deploy their infrastructure at one time and the service starts at one date. However,
WiFi service is assumed to be gradually deployed in a one hot-spot-at-a-time approach.
In this model, even though the overall 3G cost structure is much more complicated, the
authors confine cost of 3G provider in the access layer only (i.e., base station upgrade and
T1 access line), which is equivalent to the cost of WiFi.
For the 3G provider, the authors assume that it already has an existing infrastructure
for the 2.5G wireless data service. The providers also have an advantage for using exiting
base stations, and it could share the allotted spectrum band for its 2G voice service.
A low-density area could be covered by one 3G base station, while in the high-density
area, 100 3G base stations are needed for coverage. In both cases, upgrading is assumed
to be done at one time. The upgrade cost of a base station to 3G from 2.5G is expected to
be about $250,000 (Bakhshi, 2001). The T1 internet access line is assumed from the base
station, with cost calculated for 12 months.
Analysis of mobile broadband competition 593
For the WiFi provider, the equipment cost for the WiFi service is relatively low:
$1000 per access point. The authors assume the average number of access points
per hot-spot is 5 (in high-density areas) and 1 (in low-density areas). For authentication
and security, WiFi providers need a VPN gateway ($220013). They also incur a monthly
space rental fee for the access point (high-density, $40014/low-density, $100) and a
monthly T1 internet connection fee for every hot-spot (high-density, $30015/low-density,
$500). In summary, the annual cost of the WiFi provider is assumed to be approximately
$15,60016 (in high-density areas)/$10,40017 (in low-density areas). Unlike 3G providers,
WiFi providers do not have to deploy the WiFi networks all at once. Their deployment
strategy is assumed to be one hot-spot-at-a-time as dictated by demand. Table 6 shows
the cost comparison between the two providers.
3G WiFi
Density High Low High Low
Fixed 100 × 250,000 1 × $250,000 5 × $1000/access point 1 × $1000/access point
equipment cost $2200/VPN gateway $2200/VPN gateway
Monthly $300/T1 $500/T1 $300/T1, $500/T1,
variable cost $400/space rental fee $100/space rental fee
Find the best response from the above 3G provider’s responses by comparing
calculated profits
}
(2)Find WiFi provider’s response:
Repeat P3G = $50 ~ $100 {
Repeat PWF = $20 ~ $80 {
Calculate profit (response) of WiFi provider
}
Find the best response from the above WiFi provider’s responses by comparing
calculated profits
}
(3)Compare the best responses of each provider and find an equilibrium price set. For
example, (P*3G, PWF) = ($50, $20) and (P*WF, P3G) = ($20, $50), then equilibrium
price set is (P*3G, P*WF) = ($50, $20).
Figure 1 Equilibrium profits of low density area: (a) 3G and (b) WiFi (see online version
for colours)
(a)
(b)
Figure 2 Equilibrium profits of high density area: (a) 3G and (b) WiFi (see online version
for colours)
(a)
(b)
To find which provider has a higher equilibrium profit, the authors calculate
Profit ∗3G − ProfitWF
∗
. The following table and graph show
For example, in the table of Figure 3, when αWF = 10% and p-ratio = 10%, in the low
density area 3G is a market winner. In the graph, the height of the bar shows relative
market strength. If the bar in the graph is upward, the profit of 3G is higher than that
of WiFi, and if the bar is downward, it shows that the profit of WiFi is higher.
Among 100 cases based on the αWF and p-ratio, WiFi has a better market performance
in 25 cases (low density) and 19 cases (high density) respectively. In both density areas,
low αWF value (10%) and high p-ratio (100%) gives the WiFi provider maximum
relative market strength, while high αWF value (100%) and low p-ratio (100%) gives the
3G provider maximum relative market strength.
7 Discussion
As we mentioned in the introduction section, for the last several years some
US cities have deployed city-wide WiFi networks, and some have plans to develop
WiFi based municipal wireless networks. According to Daggett (2007), many US cities
are currently developing citywide broadband networks, especially large cities such as
Philadelphia, San Francisco, Minneapolis, Boston, Houston, and Seattle. Given this level
of interest there is a strong possibility for municipal wireless to prevail in the USA as a
preferred internet access platform. There are many success stories of relatively small size
cities such as St. Anthony Village, MN (8000 populations, 2.3 mile2), Granbury,
TX (6400 populations, 10 mile2), Chaska, MN (18,000 population, 13.7 mile2) effectively
employing WiFi networks within their communities.
According to the muniwireless.com, at the end of 2006, 79 US cities had their own
municipal wireless networks, and 140 US cities planned to deploy it (MuniWireless,
2007). The authors visited 79 city websites and reviewed their operation of municipal
wireless networks. We found 69 municipal wireless services and made three categories
based on their population size. Table 8 is a summary of the 69 municipal wireless
networks.
the city’s telecommunication cost including 3G wireless service for field employees.
Scott (2005) also identified that it was expected to reduce the ‘digital divide’ (the gap
between those who can afford access to the internet and those who cannot) by providing
the pressure of price competition to other wireless broadband providers. Once the WiFi
municipal network service is provided, it is expected to compete with existing
3G network service. However, the project has been stopped. In 2008, many municipal
wireless projects, especially big cities, have been delayed or cancelled.
In our game model analysis, comparing outputs between low density area and high
density area, 3G has more winning cases in the high density area than in the low
density area. Table 9 shows the number of winning cases of 3G and WiFi in the
both density. In Figures 3 and 4, when the coverage percentage is 60% through 100%,
3G is always a winner regardless of p-ratio. Therefore, the municipal wireless network of
large city with a large population like Philadelphia (1.5 million populations, 135 mile2)
has a strong possibility of being beaten by the commercial 3G network.
3G WiFi
Low density 75/100 25/100
High density 81/100 19/100
Figure 3 (a) Higher equilibrium profit provider of low density area and (b) Relative market
strength (Pf(3G) – Pf(WF)) of low density area (see online version for colours)
(a)
(b)
P: Lower left
α: Upper right
Analysis of mobile broadband competition 599
Figure 4 (a) Higher equilibrium profit provider of high density area and (b) Relative market
strength (Pf(3G) – Pf(WF)) of high density area (see online version for colours)
(a)
(b)
P: Lower left
α: Upper right
8 Conclusion
3G and WiFi are two main technologies in the mobile communications industry.
Since 3G is a WAN based technology and WiFi is a LAN based technology, they have
been complementary services. With the advent of the WiFi based municipal wireless
network, there is a strong possibility for WiFi to compete with 3G network in city areas
and potentially become a substitute. The purpose of municipal wireless network is to
reduce the 3G wireless expense of a city’s field workers as well as to provide a relatively
inexpensive broadband internet access to city’s resident and business in the city.
In this paper, the authors used a game theoretic competition model of 3G and WiFi
and demonstrated which technology would be a winner in cases of different market
penetration rates and WiFi coverage percentages. The optimal pricing of 3G and WiFi in
this model is ($50, $20), which is lower than the current actual price in the industry
($60, $30). In this analysis, we found that 3G network is more profitable as
WiFi coverage percentage increases, and 3G is more favourable in the high density areas.
WiFi has an advantage when the market has a high penetration rate and a low coverage
area. Currently, municipal wireless networks not being active in the USA’s big cities and
3G service being in the early stages, has resulted in little competition between the two.
600 S. Shin and M.B.H. Weiss
However, as the market matures and competition increases between two network service
providers, the results of this paper will help to serve as a guideline for future mobile
internet access service.
References
Bakhshi, S.K. (2001) 3G Infrastructure Vendor Profile: Nortel Networks, IDC Report.
Broadcom (2003) IEEE802.11g: The New Mainstream Wireless LAN Standard, Available at
http://www.broadcom.com/collateral/wp/802.11g-WP104-R.pdf
Daggett, B. (2007) Localizing the Internet: Five Ways Public Ownership Solves the US Broadband
Problem, Available at http://www.newrules.org/info/5ways.pdf
Dekleva, S., Shim, J.P., Varshney, U. and Knoerzer, G. (2007) ‘Evolution and emerging issues
in mobile wireless networks’, Communications of the ACM, Vol. 50, No. 6, pp.38–43.
Henkel, D., Houchaime, N., Locatelli, N. and Singh, S. (2002) The Impact of Emerging WLAN on
Incumbent Cellular Service Providers in the USA, Available at http://www.
mobilein.com/3GWLANFuture1.pdf
Lehr, W. and McKnight, L.W. (2003) ‘Wireless internet access: 3G vs. WiFi?’,
Telecommunications Policy, Vol. 27, pp.351–370.
Mossberg, W. (2006) ‘Cingular joins rivals with fast, reliable wireless broadband’,
Wall Street Journal, 19 January, 2006 Edition, Available at http://ptech.allthingsd.com/
20060119/cingular-wireless-broadband/
MuniWireless (2007) Muniwireless.com 29 December 2006 List of US Cities and Regions,
Available at http://muniwireless.com/reports/docs/Dec-29-2006summary.pdf
Odlyzko, A. and Tilly, B. (2005) A Refutation of Metcalfe’s Law and a Better Estimate for the
Value of Networks and Network Interconnections, Available at http://www.dtc.umn.edu/
~odlyzko/doc/metcalfe.pdf
Qualcomm (2004) 1xEV-DO White Paper, Available at http://www.qualcomm.com/technology/
1xev-do/published.html
Scott, E. (2005) Wireless Philadelphia, Kennedy School of Government Case Program
No. C14-05-1824.
Shin, S., Weiss, M.B. and Tucci, J. (2007) ‘Rural internet access: over-subscription strategies,
regulation and equilibrium’, Managerial and Decision Economics, Vol. 28, pp.1–12.
Strategic Policy Research (2001) 3G Wireless Market Assessment, Available at http://www.spri.
com/pdf/reports/news/SPR_3G_Sept_2001.pdf
The Wireless Philadelphia Executive Committee (2005) Wireless Philadelphia Business
Plan, Available at http://www.phila.gov/wireless/pdfs/Wireless-Phila-Business-Plan-040305-
1245pm.pdf
Wexler, J. (2008) 2008 Wireless LAN State-of-the-Market Report, Available at http://www.
webtorials.com
Notes
1
http://www.ctia.org
2
Verizon wireless was merged with Alltel, the fifth wireless operator in USA in 2008.
3
AT&T Mobile was merged with Cingular in 2006.
4
CDMA2000 is the 3rd Generation solution, which is an evolution of a 2G CDMA standard.
5
W-CDMA is also called UMTS, which is an evolution of a 2G GSM standard.
6
GPRS (AT&T, T-Mobile) and 1x-RTT (Verizon Wireless, Sprint).
Analysis of mobile broadband competition 601
7
T-Mobile was the last company in the four providers that upgraded its wireless network into 3G.
8
General packet radio services, known as a GSM based 2.5G technology.
9
http://selfcare.hotspot.t-mobile.com/locations/viewLocationMap.do
10
http://www.wayport.net/NewsReleases.aspx?id=1980
11
2.5 million/25,000/cell = 100 cells.
12
10,000 = 78.5 mile2/(0.052 × 3.14).
13
The cost of Cisco VPN concentrator 3000.
14
This number is based on the Manhattan, NYC market.
15
www.bandwidth.com, price of bandwidth is based on the distance to a collocation.
Usually, a rural area has a longer distance to the collocation than a metro area.
16
$15,600 = $2200 + 5 × $1000 + 12 × ($300+$400).
17
$10,400 = $2200 + 1 × $1000 + 12 × ($500 + $100).