Elements of a Wireless Network Architecture

Elements of a Wireless Network Architecture
It is useful to consider the elements of a wireless network in four categories: services,
infrastructure, protocols, and network engineering.
Services. From the perspective of the user of the network, the principal aspect of
the network is the service or set of services the network is designed to support. In
fact, the various industry efforts that lead to interoperability standards invariably begin
with agreements among participants as to the array of services to be provided by the
intended network standard, and considerable attention is given to the detailed features
of those services and the specific ways in which the user’s equipment will interact with
the network in the operation of the service. Of course, the basic types of services are
voice and data services.
In some networks, the voice services might comprise a menu of selectable digital
data rates, the higher data rates providing a higher received voice quality at the cost
of higher bandwidth requirement, accompanied by an appropriate tariff differential.
Data services may be provided in various forms, the simplest, called data-bearer
service, being simple transport of data with minimal specification of data format at the
mobile data port. Data service offerings might include a choice of transparent (T) or
nontransparent (NT) data in either synchronous (clock-driven) or nonsynchronous
(start/stop character-driven) formats. Transparent data service will employ forwarderror
correction coding at a fixed transmission rate in the channel. Nontransparent
service will employ error-detection coding and retransmission of faulty data blocks
so as to ensure greater accuracy in the delivered user data. Other options might include circuit-switched (connection-oriented) data versus packet (connectionless) service.
Other, application-specific data services, such as Group-3 Facsimile service, will
typically be offered with a set of optional data rates.
Short messaging service (SMS) is available in many cellular networks for transmission
and reception of short text messages displayed on a small screen. The SMS
messages are embedded in the control channels of the cellular network, which enables
rapid delivery. A service that is growing rapidly, called text messaging, is built on
SMS. As the demand for wireless multimedia service grows, data services are being
provided at increasingly high data rates.
System Infrastructure. Provisioning of the various services in a wireless network
in turn places requirements on hardware and software that must be included in the
elements connecting the wireless service customer with the fixed networks. We need
to consider two categories of system elements: the mobile terminal and the fixed
wireless infrastructure that makes connection with the fixed network.
The mobile terminal is the user’s device for sending and receiving signals over a
wireless link. For a user requiring only basic voice service, the mobile terminal is the
familiar cellular phone, nowadays probably a CDMA digital phone in the United States
or a GSM phone in Europe and many other regions of the world. Many cellular phones
are designed with a standardized data port for connecting to a portable computer or
other data terminal. In supporting data connectivity, the cellular phone is functioning
as a wireless modem interfacing baseband data (e.g., ASCII-formatted) with the
wireless network. Currently, this wireless modem function is typically implemented
in a circuit card to be plugged into a socket on a laptop computer, or even a card
already mounted inside the laptop. As wireless networks evolve to support increasing
capability for multimedia transmission, a variety of new mobile devices are appearing
in the marketplace to support sending and receiving multimedia images.
Signals from the mobile user terminal arrive at an antenna that provides an RF
interface to the fixed wireless infrastructure, and that infrastructure in turn provides
an interface to the fixed wired infrastructure. In the case of cellular systems, the fixed
wired infrastructure will typically be a public-switched telephone network (PSTN) or
a public-switched data network (PSDN). In the case of WLAN systems, the fixed
network will typically be a wired Ethernet LAN in an office building, office complex,
or university campus.
In the case of cellular systems, the fixed wireless infrastructure includes antennas,
radio base stations (BSs), mobile switching centers (MSCs), and terrestrial lines (typically,
coaxial cable or optical fiber) to make connections among BSs and MSCs as well
as between MSCs and the PSTN. The fixed wireless infrastructure will also include
computers and a variety of instrumentation needed for operation and maintenance of
the cellular network. All of the equipment and software in place, from the antennas to
the PSTN connections, will be owned and operated by the cellular service provider.
Currently, a cellular service company might have to deploy 50 to 100 BSs to provide
satisfactory signal coverage over a major metropolitan area.
Functional partitioning between network equipment elements may vary from one
manufacturer’s equipment to another’s, but in current cellular networks, the BS will
typically include not only RF transmission and reception equipment but also speech
coder/decoders (codecs). In such a configuration, all transmissions between the BS
and the PSTN are in digital form. In such a configuration, the BS will also typically include interworking functions (IWFs), also called modem emulators, to modulate and
demodulate the data streams in support of wireless data services. The MSCs include
mobile-aware switches that provide for the setup and routing of call connections to
and from mobile terminals and also handle the hand-off of call connections from one
BS to another as mobile users move about the cellular service area. The MSCs also
include the other hardware and software elements that are needed for mobile network
operation, maintenance, and troubleshooting.
Wired Backbones for Wireless Networks. Since wireless networks depend heavily
on the wired infrastructures to which they connect, in this section we provide a brief
overview of the important wired infrastructures. The most commonly used wired infrastructures
for wireless networks are PSTN, Internet, and hybrid fiber coax (HFC),
originally designed for voice, data, and cable TV distributions applications, respectively.
Figure 1.4 provides an overall picture of these three networks and how they
relate to other wired and wireless networks. (A more detailed discussion of this topic
can be found in [Pah02a].)
The main sources of information transmitted through telecommunication devices
are voice, data, and video. Voice and video are analog in nature, whereas data traffic
is digital. The dominant voice application is telephony, that is, a bidirectional symmetric
real-time conversation. To support telephony, telephone service providers have
developed a network infrastructure that establishes a connection for a telephone call during the dialing process and disconnects it after completion of the conversation. This
network is referred to as the public switched telephone network (PSTN). As shown at
the top of Fig. 1.4, the cellular telephone infrastructure provides wireless access to the
PSTN. Another network attached to the PSTN is the private branch exchange (PBX),
a local telephone switch owned privately by a business enterprise. This private switch
allows privacy and flexibility in implementing additional services in an office environment.
The PSTN physical connection to homes is twisted-pair wiring that is also used
for broadband xDSL services. The core of the PSTN is a huge digital transmission
system that allocates a 64-kb/s channel for each direction of a telephone conversation.
Other network providers often lease the PSTN transmission facilities needed to
interconnect their nodes.
The infrastructure developed for video applications is cable television, shown in the
lower part of Fig. 1.4. This network broadcasts wideband video signals to residential
premises. A cable goes from an end office to a residential neighborhood, and all
customers are fed from the same cable. The set-top boxes leased by cable companies
provide selectivity of channels, depending on the customer’s service subscription. The
end offices, where groups of distribution cables arrive, are connected to one another
with fiber lines. For this reason, the cable TV network is also called hybrid fiber coax
(HFC). Nowadays, cable distribution is also used for broadband residential access
to Internet.
The data network infrastructure was developed for bursty data applications and
evolved into the Internet, which supports Web access, e-mail, FTP, and Telnet applications
as well as multimedia (voice, video, and data) sessions with a wide variety of
session characteristics. The middle part of Fig. 1.4 shows the Internet and its relation
to other data networks. From a user point of view, data-oriented networks are always
connected, but they use the transmission resources only when a burst of information
is to be transferred. Sessions of popular data communications applications such as
Web browsing or FTP are often asymmetric, and a short upstream request burst results
in downstream transmission of a large amount of data. Symmetric sessions such as
IP telephony over data networks (termed voice over IP, or VoIP) are also becoming
popular, providing an alternative to traditional telephony. Residential Internet access
is a logical access that is physically implemented on other media, such as cable TV
wiring or copper telephone lines. Distribution of the Internet in office areas is usually
through Ethernet local area networks (LANs). Wireless LANs in offices are usually
connected to the Internet through the wired LANs. Nowadays all other private data
networks (PDNs), such as those used by banks or airline reservation agencies, are
also connected to the Internet. The Internet also serves as the backbone for wireless
data services. protocol layering is the Open System Interconnect (OSI) seven-layer reference model,
adopted as an international standard in 1978. In the OSI model, the lowest layer,
layer 1, the physical layer, provides a physical medium for the flow of information
across a link. The highest layer in the model, layer 7, the application layer, provides
services to users of the network. In the intermediate five layers, the services provided
move progressively away from the physical medium toward network- and applicationrelated
functions.
The basic concept of protocol layering is to manage the complexity of a network
design by segmenting the system functions into a set of layers, each layer built on the
ones below it. Each protocol layer can be described as performing specific services
for the higher layers while isolating the higher layers from the details of how the
services are actually implemented. The set of rules by which information is processed
and formatted in any give layer constitutes a protocol for that layer. This assures, for
example, that two pieces of equipment performing functions in the same layer can
interoperate properly at that layer. A set of layers and their protocols is commonly
referred to as a network architecture. A list of protocols used in a chosen system, one
protocol per layer or sublayer, is referred to as a protocol stack [Gar00].
In all of the wireless networks we consider in this book, the system functions are
organized according to some version of protocol layering. From one network standard
to another, the functional segmentation into layers may be somewhat different. However,
the functional segmentation will be the same for all hardware and software
elements manufactured to each particular standard. For example, the GSM network
architecture consists of five layers: transmission, radio resource management, mobility
management, communication management, and operation, administration, and maintenance
[Mou92, Meh97, Hei99]. As a second example, the IEEE 802.11 family of
standards encompasses two layers: MAC and PHY [O’Ha99]. Traffic Engineering and Deployment. The cost of equipping and deploying a wireless
communications network can vary widely depending on the type of network and the
application for which it is intended. A WLAN might be installed in a business office
or in a university campus building for a few thousand dollars. On the other hand, a
cellular telephone network built to serve a metropolitan area might incur costs of tens
of millions of dollars. However, regardless of the wireless technology employed or
the intended application, principles of sound network engineering apply: The network
should be designed to provide good signal coverage to wireless terminals over the
intended floor, campus, or geographic service area with a reasonable expenditure of
capital for equipment and installation.
In the case of a WLAN installation, access points (sometimes called base stations)
will typically be installed on ceilings or high on walls in locations chosen to provide
unobstructed signal coverage for some set of wireless terminals, such as desktop or
laptop computers. Multiple access points will be installed to cover the total population
of wireless terminals, typically with overlapping coverage areas to avoid gaps
in coverage.
In the deployment of a cellular telephone network, the general principle of good
network engineering is the same as for a WLAN deployment—install a number of cell
sites in such a way as to provide unbroken signal coverage for mobile users over the
geographic area in which the cellular company offers service. The cost of equipping
and installing a single cell site might well be on the order of $1 or 2 million, including acquisition of real estate, and thus it is important that the cell site layout be designed
to make optimum use of capital investment.
A key element in the planning of a wireless network is a specification of the traffic
the network will be designed to handle. In the case of a WLAN design, we would
want to know the number of wireless terminals and some statistics for the amount
and type of traffic to be generated by the terminals. We would want to know the
profile of short-message traffic, long file transfers, and so on, and the frequency of
these transmissions. In other words, we would like to have a traffic model as a starting
point for planning the network. With a traffic model in hand, and a specification of the
traffic capacity of an access point, we can determine the number of access points to
be provided. Then specifying the distribution of wireless user terminals will allow us
to position the access points appropriately.
In the case of a cellular network deployment, the considerations are much the same,
but with the important difference that users are highly mobile and that communication
traffic patterns can change significantly from day to day and even from hour to hour.
Commuters caught in a traffic jam or in a severe rainstorm will generate an unusually
high volume of calls as they try to contact their co-workers or family members to revise
their schedules. Fans at a Sunday afternoon football game will generate high volumes
of communication traffic in the vicinity of the stadium. Another traffic characteristic
specific to the cellular case is the relatively high frequency of call handoffs as users
move about the service area. This contrasts with the typical relatively less frequent
handoffs experienced in the WLAN environment. Thus, the efficient engineering of a
cellular network must take account not only of average statistics of generated traffic
but also of the potentially high variability of the traffic. Once again, a traffic model
is needed, and the traffic model in the cellular case is likely to be considerably more
complex than in the WLAN case. We shall have more to say about wireless network
deployment in subsequent sections.