Biblio5

11.6 LAN ACCESS PROTOCOLS

311

Figure 11.7 CSMA/ CD LAN relationship to the OSI model as well as the functional blocks required. (From IEEE 802.3 [Ref. 3]; Courtesy of the IEEE, New York.)

discussed here is based on IEEE 802.3, which closely resembles Ethernet with changes in frame structure and an expanded set of physical layer options. Figure 11.7 relates CSMA/ CD protocol layers to the conventional OSI reference model described in the previous chapter. The figure also identifies acronyms that we use in this description. The bit rates generally encompassed are 10 Mbps and 20 Mbps. We will also briefly cover an IEEE 802.3 subset standard for 100-Mbps operation. The model used in this present discussion covers the 10-Mbps data rate. There is also a 100-Mbps CSMA/ CD option, which is described in Section 11.6.2.2. There will also be a 1000-Mbps variance of this popular MAC protocol.

The medium described here is coaxial cable. However, there is a trend to use twisted wire pair. The LAN station connects to the cable by means of a medium access unit (MAU). This connects through an attachment unit interface (AUI) to the data terminal equipment (DTE). As illustrated in Figure 11.7, the DTE consists of the physical signaling sublayer (PLS), the medium access control (MAC), and the logical link control (LLC). The PLS is responsible for transferring bits between the MAC and the cable. It uses differential Manchester coding for the data transfer. With such coding the binary 0 has a transition from high to low at midcell, while the binary 1 has the opposite transition.

CSMA/ CD LAN systems (or Ethernet) are probably the most widely used type of LANs worldwide. We would say that this is due to their relatively low cost to implement and maintain and to their simplicity. The down side is that their efficiency starts to drop off radically as the number of users increases, as well as increased user activity. Thus

312 ENTERPRISE NETWORKS I

the frequency of collisions and backoffs increases to a point that throughput can drop to zero. Some users argue that efficiency starts to drop off at around 30% capacity, while others argue that that point is nearer 50%. We will discuss ways to mitigate this problem in our coverage of bridges.

11.6.2.1.1 System Operation

Transmission without contention. A MAC frame is generated from data from the LLC sublayer. This frame is handed to the transmit media access management component of the MAC sublayer for transmission. To avoid contention with other traffic on the medium, the transmit medium access management monitors the carrier sense signal provided by the physical layer signaling (PLS) component. When the medium is clear, frame transmission is initiated through the PLS interface. When the transmission has been completed without contention (a collision event), the MAC sublayer informs the LLC and awaits the next request for frame transmission.

Reception without contention. At each receiving station, the arrival of a frame is first detected by the PLS, which responds by synchronizing with the incoming preamble and by turning on the carrier sense signal. The PLS passes the received bits up to the MAC sublayer where the leading bits are discarded, up to and including the end of the preamble and start frame delimiter (SFD). In this period the receive media access management component of the MAC sublayer has detected the carrier sense and is waiting for the incoming bits to be delivered. As long as the carrier sense is on, the receive media access management collects bits from the PLS. Once the carrier sense signal has been removed, the frame is truncated at an octet boundary, if required, and then passed to the receive data decapsulation for processing.

It is in receive data decapsulation where the destination address is checked to determine if this frame is destined for this particular LAN station. If it is, the destination address (DA) and source address (SA) and the LLC data unit are passed to the LLC sublayer. It also passes along the appropriate status code indicating that reception is complete or reception too long. It also checks for invalid MAC frames by inspecting the frame check sequence (FCS) to detect any damage to the frame en route, as well as by checking for proper octet-boundary alignment of the end of frame.

Collision handling. A collision is caused by multiple stations attempting to transmit at the same time, in spite of their attempts to avoid this by deferring. A given station can experience a collision during the initial part of its transmission (the collision window) before its transmitted signal has had time to propagate to all stations on the CSMA/ CD medium. Once the collision window has passed, a transmitting station is said to have acquired the medium. Once all stations have noticed that there is a signal on the medium (by way of carrier sense), they defer to it by not transmitting, avoiding any chance of subsequent collision. The time to acquire the medium is thus based on the round-trip propagation time of the physical layer whose elements include the PLS, the physical medium attachment (PMA), and the physical medium itself.

In the event of collision, the transmitting station’s physical layer notices a marked increase in standing waves on the medium2 and turns on the collision detect (CD) signal. The collision-handling process now starts. First, the transmit media access management enforces the collision by transmitting a bit sequence called jam. This “jam,” specified as 32 bits long, ensures that all stations involved in the collision are aware that a collision has occurred. After the jam has been sent, the transmit media access

2Standing waves are called “interference” in the ISO/ IEC reference standard. Also, the signal swings because two signals are reinforcing and then nulling out each other.

11.6 LAN ACCESS PROTOCOLS

313

Figure 11.8 MAC frame format (Ref. 3).

management component terminates the transmission and schedules another transmission attempt after a randomly selected time interval. Retransmission is attempted again in the face of repeated collisions. If, on this second attempt, another collision occurs, the transmit media access management attempts to reduce the medium’s load by backing off, meaning it voluntarily delays its own retransmissions to reduce the load on the medium. This is accomplished by expanding the interval from which the random retransmission time is selected on each successive transmission attempt. Eventually, either the transmission succeeds or the attempt is abandoned on the assumption that the medium has failed or has become overloaded.

The MAC frame is shown in Figure 11.8. There are eight fields in the frame: preamble, SFD, the addresses of the frame’s destination(s) and source, a length field to indicate the length of the following field containing the LLC data, a field that contains padding (PAD) if required,3 and the FCS for error detection. All eight fields are of fixed size except the LLC data and PAD fields, which may contain any integer number of octets (bytes) between the minimum and maximum values determined by a specific implementation.

The minimum and maximum frame size limits refer to that portion of the frame from the destination address field through the frame check sequence field, inclusive. The default maximum frame size is 1518 octets; the minimum size is 64 octets.

The preamble field is 7 octets in length and is used so that the receive PLS can synchronize to the transmitted symbol stream. The SFD is the binary sequence 10101011. It follows the preamble and delimits the start of frame.

There are two address fields: the source address and the destination address. The address field length is an implementation decision. It may be 16 or 48 bits long. In either field length, the first bit specifies whether the address is an individual address (bit set to 0) or group address (bit set to 1). In the 48-bit address field, the second bit specifies whether the address is globally administered (bit set to 0) or locally administered (bit set to 1). For broadcast address, the bit is set to 1.

The length field is 2 octets long and indicates the number of LLC data octets in the data field. If the value is less than the minimum required for proper operation of the protocol, a PAD field (sequence of octets) is appended at the end of the data field and prior to the FCS field.4 The length field is transmitted and received with the high-order octet first.

The data (LLC data) field contains a sequence of octets that is fully transparent in that any arbitrary sequence of octet values may appear in the data field up to the maximum number specified by the implementation of this standard that is used. The maximum

3Padding means the adding of dummy octets (bytes) to meet minimum frame length requirements. 4Minimum frame length is 64 octets.

314 ENTERPRISE NETWORKS I

size of the data field supplied by the LLC is determined by the maximum frame size and address size parameters of a particular implementation.

The FCS field contains four octets (32 bits) CRC value. This value is computed as a function of the contents of the source address, destination address, length, LLC data, and pad—that is, all fields except the preamble, SFD, and FCS. The encoding is defined by the following generating polynomial:

G(x) c x 32 + x 26 + x 23 + x 22 + x 16 + x 12 + x 11 + x 10 + x 8 + x 7 + x 5 + x 4 + x 2 + x + 1.

An invalid MAC frame meets at least one of the following conditions:

1. The frame length is inconsistent with the length field. 2. It is not an integral number of octets in length.

3. The bits of the received frame (exclusive of the FCS itself) do not generate a CRC value identical to the one received. An invalid MAC frame is not passed to the LLC.

The minimum frame size is 512 bits for the 10-Mbps data rate (Ref. 3). This requires a data field of either 46 or 54 octets, depending on the size of the address field used. The minimum frame size is based on the slot time, which for the 10-Mbps data rate is 512 bit times. Slot time is the major parameter controlling the dynamics of collision handling and it is:

•An upper bound on the acquisition time of the medium;

•An upper bound on the length of a frame fragment generated by a collision;

•The scheduling quantum for retransmission.

To fulfill all three functions, the slot time must be larger than the sum of the physical round-trip propagation time and the MAC sublayer jam time. The propagation time for a 500-m segment of 50-Q coaxial cable is 2165 ns, assuming that the velocity of propagation of this medium is 0.77 × 300 × 106 m/ s (Ref. 3).

11.6.2.12 Transmission Requirements

System model. Propagation time is critical for the CSMA/ CD access method. The major contributor to propagation time is the coaxial cable and its length. The characteristic impedance of the coaxial cable is 50 Q ± 2 Q . The attenuation of a 500-m (1640-ft) segment of the cable should not exceed 8.5 dB (17 db/ km) measured with a 10-MHz sine wave. The velocity of propagation is 0.77c.5 The referenced maximum propagation times were derived from the physical configuration model described here. The maximum configuration is as follows:

1. A trunk coaxial cable, terminated in its characteristic impedance at each end, constitutes a coax segment. A coax segment may contain a maximum of 500 m of coaxial cable and a maximum of 100 MAUs. The propagation velocity of the coaxial cable is assumed to be 0.77c minimum (c c 300,000 km/ s). The maximum end-to-end propagation delay for a coax segment is 2165 ns.

5Where c c velocity of light in a vacuum.

11.6 LAN ACCESS PROTOCOLS

315

2. A point-to-point link constitutes a link segment. A link segment may contain a maximum end-to-end propagation delay of 2570 ns and shall terminate in a repeater set at each end. It is not permitted to connect stations to a link segment.

3. Repeater sets are required for segment interconnection. Repeater sets occupy MAU positions on coax segments and count toward the maximum number of MAUs on a coax segment. Repeater sets may be located in any MAU position on a coax segment but shall only be located at the ends of a link segment.

4. The maximum length, between driver and receivers, of an AUI cable is 50 m. The propagation velocity of the AUI cable is assumed to be 0.65c minimum. The maximum allowable end-to-end delay for the AUI cable is 257 ns.

5. The maximum transmission path permitted between any two stations is five segments, four repeater sets (including optional AUIs), two MAUs, and two AUIs. Of the five segments, a maximum of three may be coax segments; the remainder are link segments.

The maximum transmission path consists of 5 segments, 4 repeater sets (with AUIs), 2 MAUs, and 2 AUIs, as shown in Figure 11.9. If there are two link segments on the transmission path, there may be a maximum of three coaxial cable segments on that path. If there are no link segments on a transmission path, there may be a maximum of three coaxial cable segments on that path given current repeater technology. Figure 11.10 shows a large system with maximum length transmission paths. It also shows the application of link segments versus coaxial cable segments. The bitter ends of coaxial cable segments are terminated with the coaxial cable characteristic impedance. The coaxial cable segments are marked at 2.5-m intervals. MAUs should only be attached at these 2.5-m interval points. This assures nonalignment at fractional wavelength boundaries.

11.6.2.2 CSMA/ CD at 100 Mbps

11.6.2.2.1 Introduction. The demand for greater information capacity and higher transmission data rates has brought about the development of a 100-Mbps CSMA/ CD option for the enterprise network. There are several versions. The version presented here is based on IEEE Std. 802.3u-1995. The MAC and LLC are identical to those discussed in Section 11.6.2.1. To make the MAC and LLC compatible with 100-Mbps operation a new interface is provided between the PHY layer and the MAC sublayer. It consists of the reconciliation sublayer (RS) and the medium independent interface

(MII). The PHY layer has been reconfigured for 100-Mbps operation on several optional transmission media configurations. The relationship among the LLC, MAC, reconciliation, MII, and PHY components is illustrated in Figure 11.10. The generic designation

Figure 11.9 Maximum transmission path.

316

Figure 11.10 Architectural positioning of 100BASE-T. (From Ref. 4, Figure 21-1, reprinted with permission of the IEEE.)

11.6 LAN ACCESS PROTOCOLS

317

for 100-Mbps CSMA/ CD is 100BASE-T (Ref. 4), which is available in several configurations:

•100BASE-T4 uses four pair of Category 3, 4, or 5 UTP (unshielded twisted pair) balanced cable.

•100BASE-TX uses two pair of Category 5 UTP balanced cable or 150-Q STP (shielded twisted pair) balanced cable.

•100BASE-FX uses two multimode fibers employing FDDI physical layer.

100BASE-T extends the bit rate of 10BASE-T to 100 Mbps. Of course, the bit rates are faster, the bit period is shorter and the frame transmission times are reduced, and the cable delay budgets are smaller, all in proportion to the change in bit rate.

11.6.2.2.2 Reconciliation Sublayer (RS) and Media Independent Interface (MII).

The purpose of this interface is to provide a simple and easy way to implement interconnection between the MAC sublayer described in Section 11.6.2.1 and the PHY (physical layer entity), and between the PHY and station-management entities. The interface has the following characteristics:

•Supports both 10-Mbps and 100-Mbps operation;

•Data and delimiters are synchronous to clock reference;

•Provides independent four-bit-wide transmit and receive paths6

•Provides simple management interface;

•Capable of driving a limited length of shielded cable. The major concepts of the MII and RS are:

•Each direction of data transfer is serviced with seven (making a total of 14) signals including: data (a 4-bit bundle), delimiter, error, and clocks.7

•Two media status symbols are used: one indicates the presence of carrier and the other indicates the occurrence of a collision.

•A management interface, composed of two signals, provides access to management parameters and services.

•The RS maps the signal set provided at the MII to the PLS service definition.

11.6.2.2.3 Physical Coding Sublayer (PCS), Physical Medium Attachment (PMA) Sublayer, and Baseband Medium Type 100BASE-T4. The objectives of 100BASE-T4 are:

• To support CSMA/ CD MAC (see Section 11.6.2.1);

•To support 100BASE-T MII, repeater and optional autonegotiation;

•To provide a 100-Mbps data rate at the MII;

6“4-bit wide” is called a nibble, which is the unit of data exchange on the MII. Of course an octet consists of two consecutive nibbles.

7Delimiter: a bit, character, or set of characters used to denote the beginning or end of a group of related bits, characters, words, or statements. Synonym: separator.

318 ENTERPRISE NETWORKS I

Figure 11.11 Use of wire pairs. Tx c transmit, Rx c receive, BI c bidirectional. (From Ref. 4, Figure 23-3, reprinted with permission from the IEEE.)

•To provide for operation over UTP Category 3, 4, or 5 cable, installed as horizontal runs at distances up to 100 m (328 feet);

•To allow for a nominal network extension of 200 m (656 ft), including

•UTP links of 100 m; and

•Two repeater networks of approximately 200-m span.

•To provide a communication channel with a mean ternary symbol error rate,8 at the PMA service interface, of less than 1 × 10−8.

The PCS transmit function accepts data nibbles from the MII, and encodes these nibbles in a 8B6T coding (described in the following) and passes the resulting ternary symbols to the PMA. In the reverse direction, the PMA conveys received ternary symbols to the PCS receive function. The PCS receive function decodes them into octets, then passes the octets one nibble at a time to the MII. The PCS also contains a PCS carrier sense function, a PCS error sense function, a PCS collision presence function, and a management interface.

The physical level communication between PHY entities (LAN station to LAN station) takes place over four twisted pairs. Figure 11.11 shows how the 4-pairs are employed.

The 100BASE-T transmission algorithm leaves one pair open for detecting carrier from the far-end (see Figure 11.11). Leaving one pair open for carrier detection greatly simplifies media access control. All collision-detection functions are accomplished using only the unidirectional pairs TX D1 and RX D2, a manner similar to 10BASE-T. This collision detection strategy leaves three pair in each direction free for data transmission, which uses an 8B6T block code, illustrated in Figure 11.12.

The 8B6T coding maps data octets into ternary symbols. Each octet is mapped into a pattern of six ternary symbols, called a 6T code group. The 6T code groups are fanned out to three independent serial channels. The effective data rate carried on each pair is one third of 100 Mbps, which is 33.333 . . . Mbps. The ternary symbol transmission rate on each pair is 6/ 8 times 33.333, or precisely 25.000 megasymbols per second.

8A ternary symbol can have one of three states (in voltage): +, 0, −. It is a means of baseband transmission.

11.6 LAN ACCESS PROTOCOLS

319

Figure 11.12 8B6T coding. (From Ref. 4, Figure 23-4, reprinted with permission from the IEEE.)

11.6.3 Token Ring

A typical token-passing ring is shown in Figure 11.13. The token ring operation, as specified in IEEE Std 802.5 (Ref. 5), has the capability of 4 Mbps or 16 Mbps data rate. A ring is formed by physically folding the medium back on itself. Each LAN station regenerates and repeats each bit and serves as a means of attaching one or more data terminals (e.g., workstations, PCs, servers) to the ring for the purpose of communicating with other devices on the network. As a traffic frame passes around the ring, all stations, in turn, copy the traffic. Only those stations included in the destination address field pass that traffic on to the appropriate users that are attached to the station. The traffic frame continues onward back to the originator, who then removes the traffic from the ring. The pass-back to the originator acts as a form of acknowledgment that the traffic had at least passed by the destination(s).

With token ring a reservation scheme is used to accommodate priority traffic. Also, one station acts as a ring monitor to ensure correct network operation. A monitor devolvement scheme to other stations is provided in case a monitor fails or drops off the ring (i.e., shuts down). Any station on the ring can become inactive (i.e., close down), and a physical by-pass is provided for this purpose.

A station gains the right to transmit frames onto the medium when it detects a token passing on the medium. Any station with traffic to transmit, on detection of the appropriate token, may capture the token by modifying it to a start-of-frame sequence and append the proper fields to transmit the first frame. At the completion of its information transfer and after appropriate checking for proper operation, the station initiates

Figure 11.13 A token-passing ring network.

320 ENTERPRISE NETWORKS I

Figure 11.14 Frame format for token ring.

a new token, which provides other stations with the opportunity to gain access to the ring. Each station has a token-holding timer that controls the maximum period of time a station may occupy the medium before passing the token on.

Figure 11.14 shows the frame format for token ring, and Figure 11.15 illustrates the token format. In these figures, the left-most bit is transmitted first. The frame format in Figure 11.14 is used for transmitting both MAC and LLC messages to destination station(s). A frame may or may not carry an INFO field.

The starting delimiter (SD) consists of the symbol sequence JK0JK000, where J and K are nondata symbols. Both frames and tokens start with the SD sequence.

The access control (AC) is 1 octet long and contains 8 bits that are formatted PPPTMRRR. The first three bits, PPP, are the priority bits. These are used to indicate the priority of a token and therefore which stations are allowed to use the token. In a system designed for multiple priority, there are eight levels of priority available, where the lowest priority is PPP c 000 and the highest is PPP c 111. The AC field contains the token bit T and the monitor bit M. If T c 0, then the frame is a token as shown in Figure 11.15. The T bit is a 1 in all other frames. The M bit is set by the monitor as part of the procedures for recovering from malfunctions. The bit is transmitted as a 0 in all frames and tokens. The active monitor inspects and modifies this bit. All other stations repeat this bit as received. The three R bits are reservation bits. These bits allow stations with high-priority PDUs to request that the next token issued be at the requested priority.

The next field in Figure 11.14 is the frame control (FC) field. It is one octet long and defines the type of frame and certain MAC and information frame functions. The first two bits in the FC are designated FF bits, and the last six bits are called ZZZZZZ bits. If FF is 00, the frame contains a MAC PDU, and if FF is 01, the frame contains an LLC PDU.

Figure 11.15 Token format for token ring.