PLANNING OF A BROADBAND WIRELESS NETWORK BASED IN

OFDM

General Considerations

Benigno Rodr´ıguez

Institute of Electrical Engineering, Universidad de la Rep´ublica, Julio Herrera y Reissig 565, Montevideo, Uruguay

Keywords:

OFDM, WiMAX, LTE, Wireless broadband networks.

Abstract:

In this paper several hints to be considered in the design of a commercial broadband wireless network are pre-

sented. The case of study is based in Worldwide Interoperability for Microwave Access (WiMAX) networks,

but the general concepts are also valid for Long Term Evolution (LTE).

The information presented here is not only valid for the task of designing a broadband wireless network, but

also it can be useful to focus research in the economically most sensitive areas.

1 INTRODUCTION

As an example, the design of a broadband wireless

network based in OFDM for 50.000 subscribers will

be analyzed.

The main concepts of New Generation Networks

(NGN) will be considered. One important concept

in NGN is the separation between services and ac-

cess technologies. In this way the development of

new services is simpliﬁed, because under this con-

cept is not necessary to be a telecommunication ex-

pert to develop services, the activity is also open for

example, to information technology experts. Also it is

an interesting opportunity for universities to develop

new telecommunication services, and to have an even

more active participation in the development of this

industry.

To enrich services also Service Control Engines

(SCEs)

are integrated to networks. They provide

trafﬁc analysis in order to apply different policies to

different customers.

2 NETWORK OVERVIEW

In next subsections the standard components of a

WiMAX or LTE network are discussed. The analy-

sis is focused in the most important aspects and in the

http://www.cisco.com/en/US/products/ps6151/index.

html

new alternatives to the development of this type of

networks.

2.1 Customer Premise Equipment

Exists an interesting variety of Customer Premise

Equipment (CPE), simple ones, with ports for Voice

over Internet (VoIP), and with VoIP and Wireless Fi-

delity (WiFi) ports. Also exist several devices (Note-

books, Netbooks, Handhelds, etc.) with the chipset

for WiMAX access embedded.

Prices are decreasing, but they are still very signif-

icant, more than 25 % of the total Capital Expenditure

(CapEx) of the network is due to CPEs (see Figure

1). Notebooks and Netbooks with WiMAX chip in-

cluded, improve this fact and simplify the situation

for operators.

Operators have applied different strategies in or-

der to reduce the cost of CPEs. In the case of Yota

(the Russian operator), they have decided to offer the

broadband access for free during some months, for

subscribers who buy their own CPEs (usually embed-

ded in Notebooks and Netbooks)

. It has two desired

consequences for the operator, ﬁrst of all the sub-

scriber pays for the CPE, and then the operator has

not to pay for it (is not part of the CapEx). To be

more precise the operator pays less for the CPE (the

money corresponding to the months during which the

service is offered for free). The second desired effect

http://www.lightreading.com/document.asp?doc id=

184325

Rodríguez B. (2010).

PLANNING OF A BROADBAND WIRELESS NETWORK BASED IN OFDM - General Considerations.

In Proceedings of the International Conference on Wireless Information Networks and Systems, pages 55-60

DOI: 10.5220/0002912100550060

 SciTePress

is that these months for free are a very good reason for

subscribers to use the WiMAX access offered by the

operator, specially if the performance is better com-

pared with the corresponding to 3G networks.

Another important consideration is that a network

designed for USB dongles to be used in indoor sce-

narios, needs more Base Stations (BSs) than one de-

signed for desktop CPEs. There are between 4 dB and

6 dB of difference between the gain of these devices.

2.2 Base Stations

The spatial diversity order is going to have an impor-

tant impact over the number of sites (BSs) in the net-

work. In Table 1, how the number of sites vary with

the spatial diversity order

, for a given area, is shown.

Table 1: Spatial diversity order versus number of sites.

Diversity Order Number of Sites

2T2R 59

2T4R 30

4T8R 18

The number of sites have an important inﬂuence

in CapEx and Operational Expenditure (OpEx). By

using a lower spatial diversity order, a reduction in

CapEx can be obtained, but later a less efﬁcient net-

work will have a higher OpEx. Thinking in a time

frame of 10 years, generally, these cheaper BSs are

not convenient (see Figure 3).

2.3 Backhaul and Backbone

A new backhaul generation with millimeter waves

(MMWs) is available in the range of 57-86 GHz. The

equipment is cheaper compared with microwaves, and

with the advantage of a lower spectral license cost.

The concept of “light licensing” is applied to MMWs.

Very directive and short range links are achievable

with this technology

. This decreases the interference

between radio links and allow a higher reuse of fre-

quencies, both effects contribute to allow a cheap li-

censing.

Also Free Space Lasers (FSLs) are an alterna-

tive to consider. In some cases they are used as re-

dundancy for MMWs links (Kamalakis et al., 2007).

Using FSLs as a backup for MMWs is quite natu-

ral, and easy considering that both technologies have

a similar range of distances and both are very di-

rective. MMWs are signiﬁcantly affected by heavy

rains (Hansryd and Eriksson, 2009) while FSL is very

http://wwwen.zte.com.cn/en/products/wireless/wimax/

200906/t20090619 172937.html

http://www.siklu.com/

sensitive to fog, then using both technologies one is

backed up by the other. These technologies allow

to develop high bit rates links at the same time that

CapEx is reduced. By using both techniques simulta-

neously, the availability can be quite acceptable.

Another way of increasing the availability is by

deploying rings of MMWs. Usually strong rains, that

reduce signiﬁcantly the performance of MMW links

are concentrated in a small area and do not affect more

than one link (hope) of the ring.

MMWs are considered as a good alternative for

achieving the capacities needed for LTE. For sure

ﬁber optic networks are also a good alternative for

backhaul, but twice more expensiveand slow to be de-

ployed due to the installation and permissions needed.

2.4 Access Service Network Gateway

Access Service Network Gateway (ASN GW) is a cen-

tral part in a WiMAX network. It is involved with mo-

bility functions and functions which are better offered

in a centralized form.

A centralized ASN GW is used to provide Macro-

mobility (mobility in the whole network). If it is only

necessary to ensure mobility in a small part of the net-

work, an architecture of distributed ASN GWs can be

used, which is a cheaper option.

In the case of LTE technology the equivalent gate-

way is the Packet Data Network Gateway (PDN GW).

2.5 NGN and Softswitch

As it was said, probably the main concept in NGN

is the separation between services and transportation

technologies. Other important characteristic of NGN,

is that these are packet based networks, with Qual-

ity of Services (QoS) and mobility. Packet based net-

works brings the problematic of Voice over IP (VoIP)

also called IP telephony. This approach to telephony

brings several opportunities but also challenges.

In this context, the Softswitch has a very important

role. Softswitch seeks to use open standards to inte-

grate NGN networks able of providing voice, data and

multimedia services. In the architecture of Softswitch

based NGN four planes are distinguished: the trans-

portation plane, control (signaling) plane, service (ap-

plication) plane and management plane (Peng et al.,

2003). Softswitch resides in control plane and focus

on the call control capabilities. There is certain inde-

pendence between planes, e.g., service plane does not

care about the implementation of call control (control

plane). The idea is that the interfaces between these

four planes could be open and standardized in order

to have a real open architecture for these networks.

WINSYS 2010 - International Conference on Wireless Information Networks and Systems

Softswitch is an excellent tool to achieve conver-

gence, being able of working with different protocols

by using generic protocols (usually ﬁve generic pro-

tocols). To achieve the coexistence between these

different protocols they are translated and encapsu-

lated to obtain a generic protocol. The beneﬁts of

these generic protocols are several, as an example,

the Softswitch can be protocol unrelated, the generic

model subsystem (a key subsystem in Softswitch) has

to cope only with ﬁve generic protocols, which makes

its task much more simple (Zhu and Liao, 2005).

Concerning to services, Sofswitch allows to in-

crease the offer of value added services. Softswitch

can also provide services by introducing speciﬁc ap-

plication servers, developing in this way a very ﬂex-

ible and scalable service architecture based in them.

Services provided by these application servers are

available to all subscribers through Softswitch. To

simplify and open the task of developing new ser-

vices, Application Programming Interfaces (APIs)

are available. Thanks to APIs, the development of

telecommunication services is not only a task for

telecommunication experts, also programmers -not so

familiar with telecommunications- can develop new

services.

2.6 Service Control Engine and Policy

Server

Service Control Engines (SCEs), also called Service

Inspection Gateways (SIGs), are devices able of mon-

itoring the trafﬁc in a network during certain time,

then some statistics about the trafﬁc in the network

can be obtained, and some decisions can be taken in

order to optimize the network use. Through these

statistics is possible to know the behavior of the cus-

tomers and then they are a good help to design new

value added services. The available services in a net-

work are going to be an important difference between

Internet Service Providers (ISPs) when the customer

is faced to chose one ISP.

SCEs are able of applying subscriber policies, al-

lowing to change the class of service delivered to one

speciﬁc subscriber, for example allowing to increase

the velocity of the connection by some time, after ask-

ing and paying for this service through a web page.

By using SCEs is also possible to apply parental con-

trol to a speciﬁc subscriber, blocking the access to

certain web pages with inconvenient content. More

knowledge about their subscribers and the opportu-

nity of delivering customized services to them em-

powers the work of an ISP. Also quota management

services are possible to deliver through the use of

SCEs (services by trafﬁc, time or modalities as “pay-

as-you-go”). Also it is possible to apply speciﬁc poli-

cies to particular trafﬁc as P2P or VoIP trafﬁc. Spam

limitation and several security improvements can be

achieved by using SCEs in the ISP network.

SCEs allows a more efﬁcient use of the network

resources at the same time that allows to customize

the service to the customer requirements; in this way

SCEs are very useful tools in NGN. The optimum use

of network resources is an important key in multiple

play networks.

3 DECISIONS TO TAKE FOR

PLANNING A WIMAX OR LTE

NETWORK

There is a set of important decisions to make for plan-

ning this kind of Networks.

3.1 Fix or Mobile Service

It is not a question of selecting a standard, for example

in WiMAX case you can use IEEE802.16eeven if you

are planning to develop a ﬁx network (e.g., to start

with less BSs). It is more a question of considering a

business plan.

There is an important difference between the num-

ber of BSs needed for a good quality mobile service,

and the number of BSs needed for a good quality ﬁx

service. It is also important to have a clear idea of

what is lost in throughput to ensure a mobile service.

Is also convenient to have in mind the lost perfor-

mance by having the CPE 4 meters away from the

window, instead of very close to it (see footnote 5).

When a mobile network is considered, omnidirec-

tional antennas in CPEs are used, because in a mobile

scenario the CPE can not be oriented in the direction

of the BS. It causes a loss of approximately 6 dB. Be-

sides this, for a mobile network a good quality han-

dover has to be provided. These facts make that for

a mobile WiMAX network, the coverage radius is ap-

proximately 600 m. In the case of a ﬁx network this

value is of approx. 1.000 m, it means that the cover-

age area for the same BS is reduced to 36 %. Then

when a mobile network is considered, 3 times more

BSs than in the case of a ﬁx network, are necessary in

order to cover the same area. For this reason is very

important to estimate in advance if the mobile ser-

vice is going to compensate the signiﬁcant increment

caused in the investment. For sure new services are

coming, and this situation can change. But also dif-

ferent phases in the development of a broadband wire-

less network can be deﬁned. The operator can start by

PLANNING OF A BROADBAND WIRELESS NETWORK BASED IN OFDM - General Considerations

deploying a ﬁx WiMAX network (based in the mobile

standard IEEE802.16e), in future by adding new BSs

to the same area can then promote and sale mobile

services, if the market accepts and pays for them.

Most of the time a nomadic service is what the

customer requires, and a real mobile service is not in-

tensively used. The cost of deploying a good quality

nomadic network is quite similar to the one required

for a ﬁx network (an amount of extra capacity has to

be considered in the case of nomadic networks to ac-

cept new users that are not currently in the area of a

given BS). A good quality nomadic network can be a

good alternative, at least to start the development of a

broadband wireless mobile network.

3.2 Using Beamforming or Not

To use Beamforming or not, is another important de-

cision to develop this kind of networks. The objec-

tive here, is not to discuss about the advantages and

disadvantages of beamforming in general, but to dis-

cuss about the performance of commercially available

beamforming implementations.

Beamforming plus Space Time Coding (STC)

achieve performance improvements between 3,75

dB and 16 dB compared with simple MIMO-STC,

(“MIMO A”). These improvements vary depending

on the velocity of the mobile terminal (16 dB for a

stationary or pedestrian and 3,75 dB for a high mobil-

ity scenario). Between different approaches to beam-

forming, the most efﬁcient beamforming technique is

“Adaptive Beamforming”.

Not all vendors agree on the convenience of de-

ploying a network with beamforming capacity, some

of them argue that the control information (for the

case of WiMAX) is delivered without beamforming,

and then the coverage area will not be extended by

using beamforming. It is not completely so, because

usually the uplink is the limitation in the coverage

area, and then, by improving the capacity of “listen-

ing the subscribers” (providing 9 dB of uplink gain)

beamforming will increase the coverage area.

By “improving” the signal, the number of BSs

(number of sites) needed to cover the same area is di-

minished, usually to 2/3 of the number of sites needed

with simple MIMO techniques. This has an important

impact in OpEx (Paolini, 2010) (the impact in CapEx,

is not always diminished, because these BSs are more

expensive than simple MIMO ones).

3.3 Number of Sites

In Figure 1, an example of the distribution of CapEx

for a WiMAX network is provided. These percent-

ages vary from network to network, but usually there

are some common points to care.

Usually the most signiﬁcant item is “Last Mile”

(BSs). In this case the best option is not to reduce the

CapEx of this item, but to ﬁnd a good compromise be-

tween the price for this item and the associated OpEx.

As it was discussed in Section 2.2, usually cheap BSs

are associated to low spatial diversity order, what re-

sults in the need of more sites and then an increased

OpEx, this effect can be observed in Figure 3.

As it was discussed in Section 2.1, CPEs play an

important role, because of the CapEx associated to

this item. Some alternatives to decrease the CapEx

for this item were already discussed.

Another important item is “Transmission” (back-

haul and backbone as it was discussed in Section 2.3).

To decrease this item, in LTE for example, the use

of MMW is considered. Being less expensive (the

associated CapEx and OpEx) and quicker to deploy,

MMW is considered as a good alternative for devel-

oping the backhaul in LTE networks.

0 1 2 3 4 5 6 7

Items

Percentage of Total CapEx

1 Last Mile

2 CPEs

3 ASN GW

4 Infraestructure and Energy

5 Transmission

6 IP Core

Figure 1: Percentages of the total Capex for each network

component.

Exist different perspectives between vendors and

operators. Usually vendors are not worried about the

OpEx to maintain a network. Then diminishing the

CapEx can be their focus, but from the side of op-

erators, both expenditures has to be considered. In

the market there are some BS options which implies

lower CapEx, but in Figure 3, how this initial advan-

tage is compensated with time can be observed, there

the comparison for two technologies of two different

vendors is shown. Some times by making an initial

investment a little bit more expensive a higher efﬁ-

ciency network (with higher spatial diversity order,

for example) can be deployed.

Considering OpEx (see Figure 2), usually “Inter-

net Access” has an important inﬂuence in the total

OpEx. Generally the operators try to obtain a good

price for these item from the international carriers,

WINSYS 2010 - International Conference on Wireless Information Networks and Systems

but there are not so many alternatives, then it is a rel-

atively simple factor to optimize. One alternative to

diminish this factor is to use “cache systems” in order

to locally store some Internet content which is often

accessed by subscribers.

Another component of OpEx is the “Spectrum

Taxes”. This is an important factor when the operator

has a mainly wireless network (wireless access, back-

haul and backbone). In this case, at least for backhaul,

MMWs represent a good alternative due to “light li-

censing”. As it was discussed previously, MMWs

are also a good alternative from the point of view of

CapEx. Besides their throughput capacity, the men-

tioned characteristics become MMWs in a good can-

didate as a backhaul technique in present and future

wireless networks. Another way of diminishing this

factor is by using BS technologies with high spatial

diversity order, in this way less sites are needed, and

then less backhaul links has to be payed.

When a wireless network is deployed, is common

to use the top of high buildings to install BSs. Usu-

ally the operator pays a rent for using this space in

the roof of the building. The “Site Rents” amount

depends mainly on the city, but an effective way of

diminishing this factor is to diminish the number of

sites, and in this way the use of high spatial diver-

sity order BSs, is a way of using less sites to offer

the same coverage. Is worth to mention that usually

the throughput capacity is not the determinant factor

in WiMAX and LTE networks, usually the number of

BSs is determined by coverage necessities more than

by capacity necessities.

Finally, energy is also diminished by using less

sites to deploy the wireless network. All this con-

siderations are important ones when the operator has

to decide between different vendors for the Last Mile

(access) technology, as it was discussed, higher spa-

tial diversity order technologies have several interest-

ing advantages.

0.5 1 1.5 2 2.5 3 3.5 4 4.5

Items

Percentage of Total OpEx

1 Site Rents

2 Energy

3 Spectrum Taxes

4 Internet Access

Figure 2: Percentages of the total OpEx for each item.

In Figure 3, the economical evolutions of total ex-

pendituresin 10 years for two network alternativesare

compared. The alternative labeled as “Low CapEx”

corresponds to a wireless network deployed with a

Last Mile technology with low spatial diversity order.

The alternative labeled as “High CapEx” uses a Last

Mile technology with higher spatial diversity order.

In this ﬁgure, the different economical efﬁciencies of

each alternative can be observed.

0 2 4 6 8 10

100

105

110

115

120

125

130

Time, years

Percentage

Low CapEx

High CapEx

Figure 3: Economical evolution for two alternatives with

different efﬁciency.

3.4 Financial Conditions

Financial conditions are another very important as-

pect to consider. This one can be the difference be-

tween a viable or not viable project. The exportation

of these technologies is an important issue for any

country, and usually banks and governments partici-

pate in the deﬁnition of the ﬁnancial conditions that

later the vendors can offer to operators. Is impor-

tant for any government to be fully conscious of the

importance of empowering its “knowledge industry”,

because the success in this area has an accumulative

effect in the economical development of a country.

3.5 WiMAX versus LTE

Usually this comparison is contaminated with partic-

ular interests, if one operator is not considering to

deploy WiMAX, it seems reasonable to discourage

WiMAX deployments, because they could be a dan-

gerous competitor for 3G for example.

Another factor that makes this comparison dif-

ﬁcult is that both technologies are in very different

stages, WiMAX is a commercial reality today, while

LTE is in the phase of tests, trials and promises.

What is clear is that both technologies have differ-

ent temporal opportunity windows. Today WiMAX is

probably the best alternative and in some years LTE

PLANNING OF A BROADBAND WIRELESS NETWORK BASED IN OFDM - General Considerations

could occupy this position. One question is: by the

start of LTE services, how many WiMAX users, sat-

isﬁed and with a medium term contract will be?. This

could affect the success of LTE in certain regions.

Probably the decision for many operators is re-

lated with legacy technologies, in the case of LTE an

important effort in backward compatibility with typi-

cal legacytechnologies as WCDMA and GSM is done

(Beming et al., 2007). It could be a reason to wait

for LTE instead of deploying a WiMAX network now.

But in the mean time, operators who take the decision

of deploying WiMAX networks will have important

advantages to offer.

LTE has a very simple architecture, but it is not

so different from WiMAX one, and both are based in

OFDM also. Concerning to LTE performance, a good

reference is the work made by the Heinrich-Hertz-

Institute in Berlin

. In this work the performance of a

LTE test bed deployed in Berlin is evaluated. Several

similarities with WiMAX coverage and performance

can be found there.

4 CONCLUSIONS

Maybe it is useful to make the network more asym-

metric, less power and gain in CPEs and more power

and gain in BSs. It makes the network more expen-

sive, but cheaper CPEs (with less gain) can be used.

Then more subscribers and services can be attended

by the network. As it was discussed, the cost of CPEs

is still a bottle-neck in the development of WiMAX

networks. A more asymmetric network will increase

the problem of having a reduced uplink, but with the

evolution and popularization of beamforming it can

be compensated. By using beamforming the BSs are

able of “listening better”, which is very good for weak

subscriber signals.

After analyzing the impact of CPEs in CapEx, is

clear that the alternative of being bought by the cus-

tomer (usually in an embedded device) is well seen

by operators, and good for the business. Thinking in

the evolution of IPv6, and the promised explosion of

devices connected to the network, WiMAX and LTE

can result an excellent alternative to access these de-

vices. This could result in an important price decrease

of WiMAX and LTE chipsets for CPEs.

NGN concepts make the development of new ser-

vices easier and open to professionals of different ar-

eas (not only telecommunication experts). This fact

and the popularity of SCEs augur the development of

an interesting set of new services in next years, increa-

http://www.nokiasiemensnetworks.com/sites/default/

ﬁles/document/LTE measurement A4 1302 0.pdf

sing the competence in this area.

Another important consideration related with the

deployment of this kind of networks, is to decide if

a mobile or simply a nomadic network is going to be

deployed, at least as the ﬁrst phase of the project. As-

suring a good quality mobile service increases the de-

ployment costs, compared with a good quality ﬁx or

nomadic service.

As it was discussed, the spatial diversity order of

BSs play an important role in OpEx. This has to be

considered at the moment of selecting a Last Mile

technology.

REFERENCES

Beming, P., Frid, L., Hall, G., Malm, P., Noren, T., Olsson,

M., and Rune, G. (2007). LTE-SAE achitecture and

performance. Ericsson Review, 3:98–104.

Hansryd, J. and Eriksson, P.-E. (2009). High-speed mobile

backhaul demonstrators. Ericsson Review, 2:10–16.

Kamalakis, T., Neokosmidis, I., Tsipouras, A., Sphicopou-

los, T., Pantazis, S., and Andrikopoulos, I. (2007). Hy-

brid free space optical/millimeter wave outdoor links

for broadband wireless access networks. In Proc. of

the 18

Annual IEEE International Symposium on

Personal, Indoor and Mobile Radio Communications

(PIMRC’07).

Paolini, M. (2010). To beamform of not to beam-

form: Can beamforming help WiMAX opera-

tors to lower deployment and operating costs?.

http://www.senzaﬁliconsulting.com/downloads/Senza

Fili Beamforming.pdf.

Peng, X., Sen, S., and Junliang, C. (2003). Research on ser-

vice solutions in softswitch based wireless networks.

In Proc. of the 14

Personal, Indoor and Mobile

Radio Communications Symposium, volume 3, pages

2744–2748, Beijing, China.

Zhu, X. and Liao, J. (2005). Prototyping softswitch with

an emphasis on protocol relationships. In Proc. of the

International Conference on Mobile Technology,

Applications and Systems, pages 5 pp.–5, Guangzhou,

China.

WINSYS 2010 - International Conference on Wireless Information Networks and Systems