MAC Frame Format

MAC Frame Format

The MAC layer receives a block of data from the LLC layer and is responsible for

performing functions related to medium access and for transmitting the data. As

with other protocol layers, MAC implements these functions, making use of a protocol

data unit at its layer; in this case, the PDU is referred to as a MAC frame.

The exact format of the MAC frame differs somewhat for the various MAC

protocols in use. In general, all of the MAC frames have a format similar to that of

Figure 12.7. The fields of this frame are

MAC control. This field contains any protocol control information needed for

the functioning of the MAC protocol. For example, a priority level could be

indicated here.

Destination MAC address. The destination physical attachment point on the

LAN for this frame.

Source MAC address. The source physical attachment point on the LAN for

this frame.

LLC. The LLC data from the next higher layer.

CRC. The cyclic redundancy check field (also known as the frame check

sequence, FCS, field). This is an error-detecting code, as we have seen in

HDLC and other data link control protocols (Lesson 6).

In most data link control protocols, the data link protocol entity is responsible

not only for detecting errors using the CRC, but for recovering from those errors by

retransmitting damaged frames. In the LAN protocol architecture, these two functions

are split between the MAC and LLC layers. The MAC layer is responsible for

detecting errors and discarding any frames that are in error. The LLC layer optionally

keeps track of which frames have been successfully received and retransmits

unsuccessful frames.

Logical Link Control

The LLC layer for LANs is similar in many respects to other link layers in common

use. Like all link layers, LLC is concerned with the transmission of a link-level protocol

data unit (PDU) between two stations, without the necessity of an intermediate

switching node. LLC has two characteristics not shared by most other link control

protocols:

1. It must support the multi-access, shared-medium nature of the link. (This differs

from a multidrop line in that there is no primary node.)

2. It is relieved of some details of link access by the MAC layer.

Addressing in LLC involves specifying the source and destination LLC users.

Typically, a user is a higher-layer protocol or a network management function in the

station. These LLC user addresses are referred to as service access points (SAPS), in

keeping with OSI terminology for the user of a protocol layer.

We look first at the services that LLC provides to a higher-level user, then at

the LLC protocol.

LLC Services

LLC specifies the mechanisms for addressing stations across the medium and for

controlling the exchange of data between two users. The operation and format of

this standard is based on HDLC. Three services are provided as alternatives for

attached devices using LLC:

Unacknowledged connectionless service. This service is a datagram-style service.

It is a very simple service that does not involve any of the flow- and

error-control mechanisms. Thus, the delivery of data is not guaranteed. However,

in most devices, there will be some higher layer of software that deals

with reliability issues.

Connection-mode service. This service is similar to that offered by HDLC. A

logical connection is set up between two users exchanging data, and flow control

and error control are provided.

Acknowledged connectionless service. This is a cross between the previous

two services. It provides that datagrams are to be acknowledged, but no prior

logical connection is set up.

Typically, a vendor will provide these services as options that the customer

can select when purchasing the equipment. Alternatively, the customer can purchase

equipment that provides two or all three services and select a specific service

based on application.

The unacknowledged connectionless service requires minimum logic and is

useful in two contexts. First, it will often be the case that higher layers of software

will provide the necessary reliability and flow-control mechanism, and it is efficient

to avoid duplicating them. For example, either TCP or the IS0 transport protocol

standard would provide the mechanisms needed to ensure that data are delivered

reliably. Second, there are instances in which the overhead of connection establishment

and maintenance is unjustified or even counterproductive: for example, data

collection activities that involve the periodic sampling of data sources, such as sensors

and automatic self-test reports from security equipment or network components.

In a monitoring application, the loss of an occasional data unit would not

cause distress, as the next report should arrive shortly. Thus, in most cases, the

unacknowledged connectionless service is the preferred option.

The connection-mode service could be used in very simple devices, such as terminal

controllers, that have little software operating above this level. In these cases,

it would provide the flow control and reliability mechanisms normally implemented

at higher layers of the communications software.

The acknowledged connectionless service is useful in several contexts. With the

connection-mode service, the logical link control software must maintain some sort

of table for each active connection, so as to keep track of the status of that connection.

If the user needs guaranteed delivery, but there are a large number of destinations

for data, then the connection-mode service may be impractical because of

the large number of tables required; an example is a process-control or automated

factory environment where a central site may need to communicate with a large

number of processors and programmable controllers; another use is the handling of

important and time-critical alarm or emergency control signals in a factory. Because

of their importance, an acknowledgment is needed so that the sender can be assured

that the signal got through. Because of the urgency of the signal, the user might not

want to take the time to first establish a logical connection and then send the data.

LLC Protocol

The basic LLC protocol is modeled after HDLC, and has similar functions and formats.

The differences between the two protocols can be summarized as follows:

1. LLC makes use of the asynchronous, balanced mode of operation of HDLC

in order to support connection-mode LLC service; this is referred to as type 2

operation. The other HDLC modes are not employed.

2. LLC supports a connectionless service using the unnumbered information

PDU; this is known as type 1 operation.

3. LLC supports an acknowledged connectionless service by using two new

unnumbered PDUs; this is known as type 3 operation.

4. LLC permits multiplexing by the use of LLC service access points (LSAPs).

All three LLC protocols employ the same PDU format (Figure 12.7), which

consists of four fields. The DSAP and SSAP fields each contain 7-bit addresses,

which specify the destination and source users of LLC. One bit of the DSAP indicates

whether the DSAP is an individual or group address. One bit of the SSAP

indicates whether the PDU is a command or response PDU. The format of the LLC

control field is identical to that of HDLC (Figure 6.10), using extended (7-bit)

sequence numbers.

For type 1 operation, which supports the unacknowledged connectionless service,

the unnumbered information (UI) PDU is used to transfer user data. There is

no acknowledgment, flow control, or error control. However, there is error detection

and discard at the MAC level.

Two other PDUs are used to support management functions associated with

all three types of operation. Both PDUs are used in the following fashion. An LLC

entity may issue a command (CIR bit = 0) XID or TEST. The receiving LLC entity

issues a corresponding XID or TEST in response. The XID PDU is used to

exchange two types of information: types of operation supported and window size.

The TEST PDU is used to conduct a loop-back test of the transmission path

between two LLC entities. Upon receipt of a TEST command PDU, the addressed

LLC entity issues a TEST response PDU as soon as possible.

With type 2 operation, a data link connection is established between two LLC

SAPS prior to data exchange. Connection establishment is attempted by the type 2

protocol in response to a request from a user. The LLC entity issues a SABME

PDU~to request a logical connection with the other LLC entity. If the connection

is accepted by the LLC user designated by the DSAP, then the destination LLC

entity returns an unnumbered acknowledgment (UA) PDU. The connection is

henceforth uniquely identified by the pair of user SAPS. If the destination LLC user

rejects the connection request, its LLC entity returns a disconnected mode (DM)

PDU.

Once the connection is established, data are exchanged using information

PDUs, as in HDLC. The information PDUs include send and receive sequence

numbers, for sequencing and flow control. The supervisory PDUs are used, as in

HDLC, for flow control and error control. Either LLC entity can terminate a logical

LLC connection by issuing a disconnect (DISC) PDU.

With type 3 operation, each transmitted PDU is acknowledged. A new (not

found in HDLC) unnumbered PDU, the Acknowledged Connectionless (AC)

Information PDU is defined. User data are sent in AC command PDUs and must

be acknowledged using an AC response PDU. To guard against lost PDUs, a 1-bit

sequence number is used. The sender alternates the use of 0 and 1 in its AC com-

mand PDU; and the receiver responds with an AC PDU with the opposite number

of the corresponding command. Only one PDU in each direction may be outstanding

at any time.

BUS/TREE LAN

This lesson provides some technical details on busltree topology LANs and MANS.

The lesson begins with an overview of the general characteristics of this topology.

The remainder of the lesson examines the use of coaxial cable and optical fiber for

implementing this topology.

Characteristics of the Bus/Tree Topology

The bus.tree topology is a multipoint configuration. That is, there are more than

two devices connected to the medium and capable of transmitting on the medium.

This situation gives rise to several design issues, the first of which is the need for a

medium access control technique.

Another design issue has to do with signal balancing. When two stations

exchange data over a link, the signal strength of the transmitter must be adjusted to

be within certain limits. The signal must be strong enough so that, after attenuation

across the medium, it meets the receiver's minimum signal-strength requirements.

It must also be strong enough to maintain an adequate signal-to-noise ratio. On the

other hand, the signal must not be so strong that it overloads the circuitry of the

transmitter, as the signal would become distorted. Although easily accomplished for

a point-to-point link, signal balancing is no easy task for a multipoint line. If any station

can transmit to any other station, then the signal balancing must be performed

for all permutations of stations taken two at a time. For n stations, that works out

to n X (n - 1) permutations. So, for a 200-station network (not a particularly large

system), 39,800 signal-strength constraints must be satisfied simultaneously; with

interdevice distances ranging from tens to thousands of meters, this would be an

extremely difficult task for any but small networks. In systems that use radiofrequency

(RF) signals, the problem is compounded because of the possibility of RF

signal interference across frequencies. A common solution is to divide the medium

into smaller segments within which pairwise balancing is possible, using amplifiers

or repeaters between segments.

Baseband Coaxial Cable

For busltree LANs, the most popular medium is coaxial cable. The two common

transmission techniques that are used on coaxial cable are baseband and broadband,

which are compared in Table 12.3. This sublesson is devoted to baseband

systems, while the next lesson discusses broadband LANs.

A baseband LAN or MAN is defined as one that uses digital signaling; that is,

the binary data to be transmitted are inserted onto the cable as a sequence of voltage

pulses, usually using Manchester or Differential Manchester encoding (see

Figure 4.2). The nature of digital signals is such that the entire frequency spectrum

of the cable is consumed. Hence, it is not possible to have multiple channels (frequency-

division multiplexing) on the cable. Transmission is bidirectional. That is, a

signal inserted at any point on the medium propagates in both directions to the

ends, where it is absorbed (Figure 12.8a). The digital signaling requires a bus topology;

unlike analog signals, digital signals cannot easily be propagated through the

branching points required for a tree topology. Baseband bus systems can extend

only a few kilometers, at most; this is because the attenuation of the signal, which

is most pronounced at higher frequencies, causes a blurring of the pulses and a

weakening of the signal to the extent that communication over larger distances is

impractical.

The original use of baseband coaxial cable for a bus LAN was the Ethernet

system, which operates at 10 Mbps. Ethernet became the basis of the IEEE 802.3

standard.

Most baseband coaxial cable systems use a special 50-ohm cable rather than

the standard CATV 75-ohm cable. These values refer to the impedance of the cable.

Roughly speaking, impedance is a measure of how much voltage must be applied to

the cable to achieve a given signal strength. For digital signals, the 50-ohm cable suffers

less intense reflections from the insertion capacitance of the taps and provides

better immunity against low-frequency electromagnetic noise, compared to 75-ohm

cable.

As with any transmission system, there are engineering trade-offs involving

data rate, cable length, number of taps, and the electrical characteristics of the cable

and the transmitlreceive components. For example, the lower the data rate, the

longer the cable can be. That statement is true for the following reason: When a signal

is propagated along a transmission medium, the integrity of the signal suffers

due to attenuation, noise, and other impairments. The longer the length of propagation,

the greater the effect, thereby increasing the probability of error. However,

at a lower data rate, the individual pulses of a digital signal last longer and can be

recovered in the presence of impairments more easily than higher-rate, shorter

pulses.

Here is one example that illustrates some of the trade-offs. The Ethernet specification

and the original IEEE 802.3 standard specified the use of 50-ohm cable

with a 0.4-inch diameter, and a data rate of 10 Mbps. With these parameters, the

maximum length of the cable is set at 500 meters. Stations attach to the cable by

means of a tap, with the distance between any two taps being a multiple of 2.5 m;

this is to ensure that reflections from adjacent taps do not add in phase [YEN83]. A

maximum of 100 taps is allowed. In IEEE jargon, this system is referred to as

10BASE5 (10 Mbps, baseband, 500-m cable length).

To provide a lower-cost system for personal computer LANs, IEEE 802.3

later added a 10BASE2 specification. Table 12.4 compares this scheme, dubbed

Cheapernet, with 10BASE5. The key change is the use of a thinner (0.25 in) cable

of the type employed in products such as public address systems. The thinner cable

is more flexible; thus, it is easier to bend around corners and bring to a workstation

rather than installing a cable in the wall and having to provide a drop cable between

the main cable and the workstation. The cable is easier to install and uses cheaper

electronics than the thicker cable. On the other hand, the thinner cable suffers

greater attenuation and lower noise resistance than the thicker cable; as a result, it

supports fewer taps over a shorter distance.

To extend the length of the network, a repeater may be used. This device

works in a somewhat different fashion than the repeater on the ring. The bus

repeater is not used as a device attachment point and is capable of transmitting in

both directions. A repeater joins two segments of cable and passes digital signals in

both directions between the two segments. A repeater is transparent to the rest of

the system; as it does no buffering, it does not logically isolate one segment from

another. So, for example, if two stations on different segments attempt to transmit

at the same time, their packets will interfere with each other (collide). To avoid

multipath interference, only one path of segments and repeaters is allowed between

any two stations. Figure 12.9 illustrates a multiple-segment baseband bus LAN.

Broadband Coaxial Cable

In the local network context, the term broadband refers to coaxial cable on which

analog signaling is used. Table 12.3 summarizes the key characteristics of broad-

band systems. As mentioned, broadband implies the use of analog signaling. FDM

is possible, as the frequency spectrum of the cable can be divided into channels or

lessons of bandwidth. Separate channels can support data traffic, video, and radio

signals. Broadband components allow splitting and joining operations; hence, both

bus and tree topologies are possible. Much greater distances-tens of kilometersare

possible with broadband compared to baseband because the analog signals that

carry the digital data can propagate greater distances before the noise and attenuation

damage the data.

Dual and Split Configurations

As with baseband, stations on a broadband LAN attach to the cable by means of a

tap. Unlike baseband, however, broadband is inherently a unidirectional medium;

the taps that are used allow signals inserted onto the medium to propagate in only

one direction. The primary reason for this is that it is unfeasible to build amplifiers

that will pass signals of one frequency in both directions. This unidimensional property

means that only those stations "downstream" from a transmitting station can

receive its signals. How, then, to achieve full connectivity?

Clearly, two data paths are needed. These paths are joined at a point on the

network known as the headend. For a bus topology, the headend is simply one end

of the bus. For a tree topology, the headend is the root of the branching tree. All

stations transmit on one path toward the headend (inbound). Signals arriving at the

headend are then propagated along a second data path away from the headend

(outbound). All stations receive on the outbound path.

Physically, two different configurations are used to implement the inbound

and outbound paths (Figure 12.8b and c). On a dual-cable configuration, the

inbound and outbound paths are separate cables, with the headend simply a passive

connector between the two. Stations send and receive on the same frequency.

By contrast, on a split configuration, the inbound and outbound paths are different

frequency bands on the same cable. Bidirectional amplifiers3 pass lower frequencies

inbound, and higher frequencies outbound. Between the inbound and outbound

frequency bands is a guardband, which carries no signals and serves merely

as a separator. The headend contains a device for converting inbound frequencies

to outbound frequencies.

The frequency-conversion device at the headend can be either an analog or

digital device. An analog device, known as a frequency translator, converts a block

of frequencies from one range to another. A digital device, known as a remodulator,

recovers the digital data from the inbound analog signal and then retransmits

the data on the outbound frequency. Thus, a remodulator provides better signal

quality by removing all of the accumulated noise and attenuation and transmitting

a cleaned-up signal.

Split systems are categorized by the frequency allocation of the two paths, as

shown in Table 12.5. Subsplit, commonly used by the cable television industry, was

designed for metropolitan area television distribution, with limited subscriber-tocentral-

office communication. It provides the easiest way to upgrade existing

one-way cable systems to two-way operation. Subsplit has limited usefulness for

local area networking because a bandwidth of only 25 MHz is available for two-way

communication. Midsplit is more suitable for LANs, because it provides a more

equitable distribution of bandwidth. However, midsplit was developed at a time

when the practical spectrum of a cable-TV cable was 300 MHz, whereas a spectrum

of 400 to 450 MHz is now available. Accordingly, a highsplit specification has been

developed to provide greater two-way bandwidth for a split cable system.

The differences between split and dual configurations are minor. The split system

is useful when a single cable plant is already installed in a building. If a large

amount of bandwidth is needed, or the need is anticipated, then a dual cable system

is indicated. Beyond these considerations, it is a matter of a trade-off between cost

and size. The single-cable system has the fixed cost of the headend remodulator or

frequency translator. The dual cable system makes use of more cable, taps, splitters,

and amplifiers. Thus, dual cable is cheaper for smaller systems, where the fixed cost

of the headend is noticeable, and single cable is cheaper for larger systems, where

incremental costs dominate.

Carrierband

There is another application of analog signaling on a LAN, known as carrierband,

or single-channel broadband. In this case, the entire spectrum of the cable is

devoted to a single transmission path for the analog signals; no frequency-division

multiplexing is possible.

Typically, a carrierband LAN has the following characteristics. Bidirectional

transmission, using a bus topology, is employed. Hence, there can be no amplifiers,

and there is no need for a headend. Although the entire spectrum is used, most of

the signal energy is concentrated at relatively low frequencies, which is an advantage,

because attenuation is less at lower frequencies.

Because the cable is dedicated to a single task, it is not necessary to take care

that the modem output be confined to a narrow bandwidth. Energy can spread over

the entire spectrum. As a result, the electronics are simple and relatively inexpensive.

Typically, some form of frequency-shift keying (FSK) is used. Carrierband

would appear to give comparable performance, at a comparable price, to baseband.

Optical Fiber Bus

Several approaches can be taken in the design of a fiber bus topology LAN or

MAN. The differences have to do with the nature of the taps into the bus and the

detailed topology.

Optical Fiber Taps

With an optical fiber bus, either an active or passive tap can be used. In the case of

an active tap (Figure 12.10a), the following steps occur:

1.Optical signal energy enters the tap from the bus.

2.Clocking information is recovered from the signal, and the signal is converted

to an electrical signal.

3.The converted signal is presented to the node and perhaps modified by the

latter.

4.The optical output (a light beam) is modulated according to the electrical signal

and launched into the bus.

In effect, the bus consists of a chain of point-to-point links, and each node acts

as a repeater. Each tap actually consists of two of these active couplers and requires

two fibers; this is because of the inherently unidirectional nature of the device of

Figure 12.10~1.

In the case of a passive tap (Figure 12.10b), the tap extracts a portion of the

optical energy from the bus for reception and it injects optical energy directly into

the medium for transmission. Thus, there is a single run of cable rather than a chain

of point-to-point links. This passive approach is equivalent to the type of taps typically

used for twisted pair and coaxial cable. Each tap must connect to the bus twice:

once for transmit and once for receive.

The electronic complexity and interface cost are drawbacks for the implementation

of the active tap. Also, each tap will add some increment of delay, just as

in the case of a ring. For passive taps, the lossy nature of pure optical taps limits the

number of devices and the length of the medium. However, the performance of

such taps has improved sufficiently in recent years so to make fiber bus networks

practical.

Optical Fiber Bus Configurations

A variety of configurations for the optical fiber bus have been proposed, all of

which fall into two categories: those that use a single bus and those that use two

buses.

Figure 12.11a shows a typical single-bus configuration, referred to as a loop

bus. The operation of this bus is essentially the same as that of the dual-bus broadband

coaxial system described earlier. Each station transmits on the bus in the

direction toward the headend, and receives on the bus in the direction away from

the headend. In addition to the two connections shown, some MAC protocols

require that each station have an additional sense tap on the inbound (toward the

headend) portion of the bus. The sense tap is able to sense the presence or absence

of light on the fiber, but it is not able to recover data.

Figure 12.11b shows the two-bus configuration. Each station attaches to both

buses and has both transmit and receive taps on both buses. On each bus, a station

may transmit only to those stations downstream from it. By using both buses, a station

may transmit to, and receive from, all other stations. A given node, however,

must know which bus to use to transmit to another node; if such information were

unavailable, all data would have to be sent out on both buses; this is the configuration

used in the IEEE 802.6 MAN, and is described in Lesson 13.

Ring LANs

Characteristics of RING LANs

A ring consists of a number of repeaters, each connected to two others by unidirectional

transmission links to form a single closed path. Data are transferred sequentially,

bit by bit, around the ring from one repeater to the next. Each repeater regenerates

and retransmits each bit.

For a ring to operate as a communication network, three functions are

required: data insertion, data reception, and data removal. These functions are provided

by the repeaters. Each repeater, in addition to serving as an active element on

the ring, serves as a device attachment point. Data insertion is accomplished by the

repeater. Data are transmitted in packets, each of which contains a destination

address field. As a packet circulates past a repeater, the address field is copied. If

the attached station recognizes the address, the remainder of the packet is copied.

Repeaters perform the data insertion and reception functions in a manner not

unlike that of taps, which serve as device attachment points on a bus or tree. Data

removal, however, is more difficult on a ring. For a bus or tree, signals inserted onto

the line propagate to the endpoints and are absorbed by terminators. Hence, shortly

after transmission ceases, the bus or tree is clean of data. However, because the ring

is a closed loop, a packet will circulate indefinitely unless it is removed. A packet

may by removed by the addressed repeater. Alternatively, each packet could be

removed by the transmitting repeater after it has made one trip around the loop.

This latter approach is more desirable because (1) it permits automatic acknowledgment

and (2) it permits multicast addressing: one packet sent simultaneously to

multiple stations.

A variety of strategies can be used for determining how and when packets are

inserted onto the ring. These strategies are, in effect, medium access control protocols,

and are discussed in Lesson 13.

The repeater, then, can be seen to have two main purposes: (1) to contribute

to the proper functioning of the ring by passing on all the data that come its way,

and (2) to provide an access point for attached stations to send and receive data.

Corresponding to these two purposes are two states (Figure 12.12): the listen state

and the transmit state.

In the listen state, each received bit is retransmitted with a small delay,

required to allow the repeater to perform required functions. Ideally, the delay

should be on the order of one bit time (the time it takes for a repeater to transmit

one complete bit onto the outgoing line). These functions are

Scan passing bit stream for pertinent patterns. Chief among these is the

address or addresses of attached stations. Another pattern, used in the token

control strategy explained later, indicates permission to transmit. Note that to

perform the scanning function, the repeater must have some knowledge of

packet format.

Copy each incoming bit and send it to the attached station, while continuing

to retransmit each bit. This will be done for each bit of each packet addressed

to this station.

Modify a bit as it passes by. In certain control strategies, bits may be modified,

for example, to indicate that the packet has been copied; this would serve as

an acknowledgment.

When a repeater's station has data to send, and when the repeater, based on

the control strategy, has permission to send, the repeater enters the transmit state.

In this state, the repeater receives bits from the station and retransmits them on its

outgoing link. During the period of transmission, bits may appear on the incoming

ring link. There are two possibilities, and they are treated differently:

The bits could be from the same packet that the repeater is still in the process

of sending. This will occur if the bit length of the ring is shorter than the

packet. In this case, the repeater passes the bits back to the station, which can

check them as a form of acknowledgment.

For some control strategies, more than one packet could be on the ring at the

same time. If the repeater, while transmitting, receives bits from a packet it

did not originate, it must buffer them to be transmitted later.

These two states, listen and transmit, are sufficient for proper ring operation.

A third state, the bypass state, is also useful. In this state, a bypass relay can be

activated so that signals propagate past the repeater with no delay other than

from medium propagation. The bypass relay affords two benefits: (1) it provides a

partial solution to the reliability problem, discussed later, and (2) it improves performance

by eliminating repeater delay for those stations that are not active on the

network.

Twisted pair, baseband coax, and fiber optic cable can all be used to provide

the repeater-to-repeater links. Broadband coax, however, could not easily be used.

Each repeater would have to be capable, asynchronously, of receiving and transmitting

data on multiple channels.

Timing Jitter

On a ring transmission medium, some form of clocking is included with the signal,

as for example with the use of Differential Manchester encoding (Lesson 4.1). As

data circulate around the ring, each repeater receives the data, and recovers the

clocking for two purposes: first, to know when to sample the incoming signal to

recover the bits of data, and second, to use the clocking for transmitting the signal

to the next repeater. This clock recovery will deviate in a random fashion from the

mid-bit transitions of the received signal for several reasons, including noise during

transmission and imperfections in the receiver circuitry; the predominant reason,

however, is delay distortion (described in Lesson 2.3). The deviation of clock recovery

is known as timing jitter.

As each repeater receives incoming data, it issues a clean signal with no distortion.

However, the timing error is not eliminated. Thus, the digital pulse width

will expand and contract in a random fashion as the signal travels around the ring

and the timing jitter accumulates. The cumulative effect of the jitter is to cause the

bit latency, or bit length, of the ring to vary. However, unless the latency of the ring

remains constant, bits will be dropped (not retransmitted) as the latency of the ring

decreases, or they will be added as the latency increases.

This timing jiiter places a limitation on the number of repeaters in a ring.

Although this limitation cannot be entirely overcome, several measures can be

taken to improve matters. In essence, two approaches are used in combination.

First, each repeater can include a phase-lock loop. This is a device that uses feedback

to minimize the deviation from one bit time to the next. Second, a buffer can

be used at one or more repeaters. The buffer is initialized to hold a certain number

of bits, and expands and contracts to keep the bit length of the ring constant. The

combination of phase-locked loops and a buffer significantly increases maximum

feasible ring size.

Potential Ring Problems

There are a number of potential problems with the ring topology: A break in any

link or the failure of a repeater disables the entire network; installation of a new

repeater to support new devices requires the identification of two nearby, topologically

adjacent repeaters; timing jitter must be dealt with; and finally, because the

ring is closed, a means is needed to remove circulating packets, with backup techniques

to guard against error.

The last problem is a protocol issue and will be discussed later. The remaining

problems can be handled by a refinement of the ring topology and will be discussed

next.

The Star-Ring Architecture

Two observations can be made about the basic ring architecture described above.

First, there is a practical limit to the number of repeaters on a ring. This limit is suggested

by the jitter, reliability, and maintenance problems just cited, and by the

accumulating delay of a large number of repeaters. A limit of a few hundred

repeaters seems reasonable. Second, the functioning of the ring does not depend on

the actual routing of the cables that link the repeaters.

These observations have led to the development of a refined ring architecture,

the star-ring, which overcomes some of the problems of the ring and allows the construction

of larger local networks.

As a first step, consider the rearrangement of a ring into a star. This is

achieved by having the interrepeater links all thread through a single site. This ring

wiring concentrator has a number of advantages. Because there is centralized access

to the signal on every link, it is a simple matter to isolate a fault. A message can be

launched into the ring and tracked to see how far it gets without mishap. A faulty

segment can be disconnected and repaired at a later time. New repeaters can easily

be added to the ring: Simply run two cables from the new repeater to the site of the

ring wiring concentration and splice into the ring.

The bypass relay associated with each repeater can be moved into the ring

wiring concentrator. The relay can automatically bypass its repeater and two links

in the event of any malfunction. A nice effect of this feature is that the transmission

path from one working repeater to the next is approximately constant; thus, the

range of signal levels to which the transmission system must automatically adapt is

much smaller.

The ring wiring concentrator permits rapid recovery from a cable or repeater

failure. Nevertheless, a single failure could, at least temporarily, disable the entire

network. Furthermore, throughput and jitter considerations still place a practical

upper limit on the number of stations in a ring, as each repeater adds an increment

of delay. Finally, in a spread-out network, a single wire concentration site dictates a

great deal of cable.

To attack these remaining problems, a LAN consisting of multiple rings connected

by bridges can be constructed.

Bus versus Ring

For the user with a large number of devices and high-capacity requirements, the bus

or tree broadband LAN seems the best suited to the requirements. For more moderate

requirements, however, the choice between a baseband bus LAN and a ring

LAN is not at all clear-cut.

The baseband bus is the simpler system. Passive taps rather than active

repeaters are used. There is no need for the complexity of bridges and ring wiring

concentrators.

The most important benefit of the ring is that it uses point-to-point communication

links, and here there are a number of implications. First, because the transmitted

signal is regenerated at each node, transmission errors are minimized and

greater distances can be covered than with baseband bus. Broadband busltree can

cover a similar range, but cascaded amplifiers can result in loss of data integrity at

high data rates. Second, the ring can accommodate optical fiber links, which provide

very high data rates and excellent electromagnetic interference (EMI) characteristics.

Finally, the electronics and maintenance of point-to-point lines are simpler

than for multipoint lines.

Star LAN

Twisted Pair Star LANs

In recent years, there has been increasing interest in the use of twisted pair as a

transmission medium for LANs. From the earliest days of commercial LAN availability,

twisted pair bus LANs have been popular. However, such LANs suffer in

comparison with a coaxial cable LAN. First of all, the apparent cost advantage of

twisted pair is not as great as it might seem, at least when a linear bus layout is used.

True, twisted pair cable is less expensive than coaxial cable. On the other hand,

much of the cost of LAN wiring is in the labor cost of installing the cable, which is

no greater for coaxial cable than for twisted pair. Secondly, coaxial cable provides

superior signal quality, and therefore can support more devices over longer distances

at higher data rates than twisted pair.

The renewed interest in twisted pair, at least in the context of busltree type

LANs, is in the use of unshielded twisted pair in a star-wiring arrangement. The reason

for the interest is that unshielded twisted pair is simply telephone wire, and virtually

all office buildings are equipped with spare twisted pairs running from wiring

closets to each office. This yields several benefits when deploying a LAN:

1. There is essentially no installation cost with unshielded twisted pair, as the

wire is already there. Coaxial cable has to be pulled. In older buildings, this

may be difficult because existing conduits may be crowded.

2. In most office buildings, it is impossible to anticipate all the locations where

network access will be needed. Because it is extravagantly expensive to run

coaxial cable to every office, a coaxial cable-based LAN will typically cover

only a portion of a building. If equipment subsequently has to be moved to an

office not covered by the LAN, significant expense is involved in extending

the LAN coverage. With telephone wire, this problem does not arise, as all

offices are covered.

The most popular approach to the use of unshielded twisted pair for a LAN is

therefore a star-wiring approach. The products on the market use a scheme suggested

by Figure 12.13, in which the central element of the star is an active element,

referred to as the hub. Each station is connected to the hub by two twisted pairs

(transmit and receive). The hub acts as a repeater: When a single station transmits,

the hub repeats the signal on the outgoing line to each station.

Note that although this scheme is physically a star, it is logically a bus: A transmission

from any one station is received by all other stations, and, if two stations

transmit at the same time, there will be a collision.

Multiple levels of hubs can be cascaded in a hierarchical configuration. Figure

12.14 illustrates a two-level configuration. There is one header hub (HHUB) and

one or more intermediate hubs (IHUB). Each hub may have a mixture of stations

and other hubs attached to it from below. This layout fits well with building wiring

practices. Typically, there is a wiring closet on each floor of an office building, and

a hub can be placed in each one. Each hub could service the stations on its floor.

Figure 12.15 shows an abstract representation of the intermediate and header

hubs. The header hub performs all the functions described previously for a singlehub

configuration. In the case of an intermediate hub, any incoming signal from

below is repeated upward to the next higher level. Any signal from above is

repeated on all lower-level outgoing lines. Thus, the logical bus characteristic is

retained: A transmission from any one station is received by all other stations, and,

if two stations transmit at the same time, there will be a collision.

Optical Fiber Star

One of the first commercially available approaches for fiber LANs was the passive

star coupler. The passive star coupler is fabricated by fusing together a number of

optical fibers. Light that is input to one of the fibers on one side of the coupler is

equally divided among, and output through, all the fibers on the other side. To form

a network, each device is connected to the coupler with two fibers, one for transmit

and one for receive (Figure 12.16). All of the transmit fibers enter the coupler on

one side, and all of the receive fibers exit on the other side. Thus, although the

arrangement is physically a star, it acts like a bus: A transmission from any one

device is received by all other devices, and if two devices transmit at the same time,

there will be a collision.

Two methods of fabrication of the star coupler have been pursued: the biconic

fused coupler and the mixing rod coupler. In the biconic fused coupler, the fibers

are bundled together and heated with an oxyhydrogen flame before being pulled

into a biconical tapered shape. That is, the rods come together into a fused mass

that tapers into a conical shape and then expands back out again. (The mixing

rod approach begins in the same fashion.) Then, the biconical taper is cut at the

waist and a cylindrical rod is inserted between the tapers and fused to the two cut

ends. This latter technique allows the use of a less-narrow waist, and it is easier to

fabricate.

Commercially available passive star couplers can support a few tens of stations

at a radial distance of up to a kilometer or more. The limitations on number

of stations and distances are imposed by the losses in the network. The attenuation

that will occur in the network consists of the following components:

0 Optical connector losses. Connectors are used to splice together cable segments

for increased length. Typical connector losses are 1.0 to 1.5 dB per

connector. A typical passive star network will have from 0 to 4 connectors in

a path from transmitter to receiver, for a total maximum attenuation of 4 to

6 dB.

0 Optical cable attenuation. Typical cable attenuation for the cable that has

been used in these systems ranges from 3 to 6 dB per kilometer.

o Optical power division in the coupler. The coupler divides the optical power

from one transmission path equally among all reception paths. Expressed in

decibels, the loss seen by any node is 10 log N, where N is the number of

nodes. For example, the effective loss in a 16-port coupler is about 12 dB.

WIRELESS LABIS

In just the past few years, wireless LANs have come to occupy a significant niche in

the local area network market. Increasingly, organizations are finding that wireless

LANs are an indispensable adjunct to traditional wired LANs, as they satisfy

requirements for mobility, relocation, ad hoc networking, and coverage of locations

difficult to wire.

As the name suggests, a wireless LAN is one that makes use of a wireless

transmission medium. Until relatively recently, wireless LANs were little used; the

reasons for this included high prices, low data rates, occupational safety concerns,

and licensing requirements. As these problems have been addressed, the popularity

of wireless LANs has grown rapidly.

In this lesson, we first look at the requirements for and advantages of wireless

LANs, and then preview the key approaches to wireless LAN implementation.

Wireless LANs Applications

[PAHL95a] lists four application areas for wireless LANs: LAN extension, crossbuilding

interconnect, nomadic access, and ad hoc networks. Let us consider each

of these in turn.

LAN Extension

Early wireless LAN products, introduced in the late 1980s, were marketed as substitutes

for traditional wired LANs. A wireless LAN saves the cost of the installation

of LAN cabling and eases the task of relocation and other modifications to

network structure. However, this motivation for wireless LANs was overtaken by

events. First, as awareness of the need for LAN became greater, architects designed

new buildings to include extensive prewiring for data applications. Second, with

advances in data transmission technology, there has been an increasing reliance on

twisted pair cabling for LANs and, in particular, Category 3 unshielded twisted pair.

Most older building are already wired with an abundance of Category 3 cable. Thus,

the use of a wireless LAN to replace wired LANs has not happened to any great

extent.

However, in a number of environments, there is a role for the wireless LAN

as an alternative to a wired LAN. Examples include buildings with large open areas,

such as manufacturing plants, stock exchange trading floors, and warehouses; historical

buildings with insufficient twisted pair and in which drilling holes for new

wiring is prohibited; and small offices where installation and maintenance of wired

LANs is not economical. In all of these cases, a wireless LAN provides an effective

and more attractive alternative. In most of these cases, an organization will also

have a wired LAN to support servers and some stationary workstations. For example,

a manufacturing facility typically has an office area that is separate from the factory

floor but which must be linked to it for networking purposes. Therefore, typically,

a wireless LAN will be linked into a wired LAN on the same premises. Thus,

this application area is referred to as LAN extension.

Figure 12.17 indicates a simple wireless LAN configuration that is typical of

many environments. There is a backbone wired LAN, such as Ethernet, that supports

servers, workstations, and one or more bridges or routers to link with other

networks. In addition there is a control module (CM) that acts as an interface to a

wireless LAN. The control module includes either bridge or router functionality to

link the wireless LAN to the backbone. In addition, it includes some sort of access

control logic, such as a polling or token-passing scheme, to regulate the access from

the end systems. Note that some of the end systems are standalone devices, such as

a workstation or a server. In addition, hubs or other user modules (UM) that control

a number of stations off a wired LAN may also be part of the wireless LAN

configuration.

The configuration of Figure 12.17 can be referred to as a single-cell wireless

LAN; all of the wireless end systems are within range of a single control module.

Another common configuration, suggested by Figure 12.18, is a multiple-cell wireless

LAN. In this case, there are multiple control modules interconnected by a wired

LAN. Each control module supports a number of wireless end systems within its

transmission range. For example, with an infrared LAN, transmission is limited to

a single room; therefore, one cell is needed for each room in an office building that

requires wireless support.

Ethernet

Cross-Building Interconnect

Another use of wireless LAN technology is to connect LANs in nearby buildings,

be they wired or wireless LANs. In this case, a point-to-point wireless link is used

between two buildings. The devices so connected are typically bridges or routers.

This single point-to-point link is not a LAN per se, but it is usual to include this

application under the heading of wireless LAN.

Nomadic Access

Nomadic access provides a wireless link between a LAN hub and a mobile data terminal

equipped with an antenna, such as a laptop computer or notepad computer.

One example of the utility of such a connection is to enable an employee returning

from a trip to transfer data from a personal portable computer to a server in the

office. Nomadic access is also useful in an extended environment such as a campus

or a business operating out of a cluster of buildings. In both of these cases, users

may move around with their portable computers and may wish access to the servers

on a wired LAN from various locations.

Ad Hoc Networking

An ad hoc network is a peer-to-peer network (no centralized server) set up temporarily

to meet some immediate need. For example, a group of employees, each

with a laptop or palmtop computer, may convene in a conference room for a business

or classroom meeting. The employees link their computers in a temporary network

just for the duration of the meeting.

Figure 12.19 suggests the differences between an ad hoc wireless LAN and a

wireless LAN that supports LAN extension and nomadic access requirements. In

the former case, the wireless LAN forms a stationary infrastructure consisting of

one or more cells with a control module for each cell. Within a cell, there may be a

number of stationary end systems. Nomadic stations can move from one cell to

another. In contrast, there is no infrastructure for an ad hoc network. Rather, a peer

collection of stations within range of each other may dynamically configure themselves

into a temporary network.

Wireless LAN Requirements

A wireless LAN must meet the same sort of requirements typical of any LAN,

including high capacity, ability to cover short distances, full connectivity among

attached stations, and broadcast capability. In addition, there are a number of

requirements specific to the wireless LAN environment. The following are among

the most important requirements for wireless LANs:

Throughput. The medium access control protocol should make as efficient use

as possible of the wireless medium to maximize capacity.

Number of nodes. Wireless LANs may need to support hundreds of nodes

across multiple cells.

Connection to backbone LAN. In most cases, interconnection with stations on

a wired backbone LAN is required. For infrastructure wireless LANs, this is

easily accomplished through the use of control modules that connect to both

types of LANs. There may also need to be accommodation for mobile users

and ad hoc wireless networks.

Service area. A typical coverage area for a wireless LAN may be up to a 300

to 1000 foot diameter.

Battery power consumption. Mobile workers use battery-powered workstations

that need to have a long battery life when used with wireless adapters.

This suggests that a MAC protocol that requires mobile nodes to constantly

monitor access points or to engage in frequent handshakes with a base station

is inappropriate.

Transmission robustness and security. Unless properly designed, a wireless

LAN may be interference-prone and easily eavesdropped upon. The design of

a wireless LAN must permit reliable transmission even in a noisy environment

and should provide some level of security from eavesdropping.

" Collocated network operation. As wireless LANs become more popular, it is

quite likely for two of them to operate in the same area or in some area where

interference between the LANs is possible. Such interference may thwart the

normal operation of a MAC algorithm and may allow unauthorized access to

a particular LAN.

" License-free operation. Users would prefer to buy and operate wireless LAN

products without having to secure a license for the frequency band used by the

LAN.

" HandoWroaming. The MAC protocol used in the wireless LAN should

enable mobile stations to move from one cell to another.

" Dynamic configuration. The MAC addressing and network management

aspects of the LAN should permit dynamic and automated addition, deletion,

and relocation of end systems without disruption to other users.

Wireless LAN Technology

Wireless LANs are generally categorized according to the transmission technique

that is used. All current wireless LAN products fall into one of the following

categories:

Infrared (IR) LANs. An individual cell of an IR LAN is limited to a single

room, as infrared light does not penetrate opaque walls.

Spread Spectrum LANs. This type of LAN makes use of spread spectrum

transmission technology. In most cases, these LANs operate in the ISM

(Industrial, Scientific, and Medical) bands, so that no FCC licensing is

required for their use in the U.S.

Narrowband Microwave. These LANs operate at microwave frequencies

but do not use spread spectrum. Some of these products operate at frequencies

that require FCC licensing, while others use one of the unlicensed ISM

bands.

Table 12.6 summarizes some of the key characteristics of these three technologies.