The Physical Layer

The Physical Layer
The easiest way to think about the Physical layer is that it deals with measurable, physical entities (like electrons, electricity, etc.). Any protocol or device that operates at the Physical layer
deals with the physical concepts of a network and knows nothing of the meaning of the individual bits that it transmits or deals with.

Physical Layer Concepts
Generally speaking, Physical layer concepts involve a network component that is tangible or measurable. For example, when a protocol at the Physical layer receives information from the upper
layers, it translates all the data into signals that can be transmitted on a transmission medium. This process is known as signal encoding
(or encoding, for short). With cable media (also called bounded media ), the protocols that operate at the Physical layer translate the bits of the data into electrical ons and offs, often using pulses of electricity or light for one or both of these states.

     Additionally, the Physical layer specifies how much of the media will be used (in other words, its signaling method ) during data transmission. If a network signal uses all available signal  requencies (or, to put it differently, the entire bandwidth), the technology is said to use baseband signaling. Most LAN technologies, such as Ethernet, use baseband signaling. On the other hand, if a signal uses only one frequency (or only part of the bandwidth), the technology is said to use broadband
signaling. This means multiple signals can be transmitted on the media simultaneously, and one or more of these channels must be “tuned” to in order for device-to-device communication to occur across them. Television signals use broadband signaling.

    Finally, the Physical layer specifies the layout of the transmission media (its topology, in other words). A physical topology describes the way the cabling is physically laid out (as opposed to a logical topology, discussed later in the section titled “The Data Link Layer”). The physical topologies include the following:

Bus
Star
Ring
Mesh


The Bus Topology
In a physical bus topology, every computer is directly connected to a common medium. A physical bus network uses one network cable that runs from one end of the network to the other.
Workstations connect at various points along this cable. The main advantage to this topology is simplicity: Only one cable is used, and a physical bus topology typically requires less cable than other physical topologies. However, a cable fault can bring down the entire network, thus making a physical bus topology the least fault tolerant of all the physical topologies.

 Figure 2.3 shows a sample physical bus network.

The Star Topology
In a physical star topology, a cable runs from each network entity to a central device. This central device (called a hub ) allows all devices to communicate as if they were all directly connected. The
main advantage to a physical star topology is its fault tolerance. If one node or cable malfunctions, the rest of the network is not affected. The hub simply won’t be able to communicate with the station attached to that port. An Ethernet 10Base-T network is one example of a network type that requires a physical star topology. Figure 2.4 shows a sample network that uses a physical star topology.

FIGURE 2 . 4 A sample physical star topology.

The Ring Topology
A physical ring topology isn’t seen much in the computer-networking world. If you do see it, it’s usually in a wide area network (WAN) environment. In a physical ring topology, every network entity connects directly to only two other network entities (the one immediately preceding it and the one immediately following it). The vulnerability of the physical ring topology to disruption of service due to the failure of a single node makes it a poor choice in most network environments. As a result, LANs only ever have implemented the ring as a logical topology, as in physical

star/logical ring Token Ring. Figure 2.5 shows a physical ring network.

FIGURE 2 . 5 A sample physical ring topology

The Mesh Topology
A physical mesh topology is another physical topology that isn’t widely used in computer networks (except in special WAN cases). In a physical mesh topology, every computer is directly
connected to every other computer in the network. The more computers there are on a mesh network, the more cables make up the network. If a mesh network has n computers, there will be n
( n– 1)/2 cables. With 10 computers, there would be 10(10– 1)/2, or 45 cables. As you can see, this topology quickly becomes unmanageable with only a few computers. Figure 2.6 shows a

sample mesh network.

Note:

It is possible to have a partial mesh network, where there are multiple connectionsbetween network entities but not between all of them. 

This reduces costs associated with leased circuits by reducing the number of circuits.

FIGURE 2 . 6 A physical mesh topology

Physical Layer Devices

Several devices operate primarily at the Physical layer of the OSI model. These devices manipulate mainly the physical aspects of a network data stream (such as the voltages, signal direction,

and signal strength). Let’s take a quick look at some of the most popular:

•  NIC
•  Transceivers
• Repeaters
• Hubs
• MAUs

The Network Interface Card (NIC)

Probably the most common component on any network is the network interface card (NIC). A NIC is the component that provides the connection between a computer’s internal bus and the

 network media. NICs come in many shapes and sizes. They vary by the type of bus connection they employ and their network media connection ports. More than any other Physical layer device, a NIC is recognized for both its layer 2 and its layer 1 personality. Think about it this way: Where in your PC is the Ethernet protocol? We know Ethernet is a layer 2 protocol, but your computer, not the NIC, is the layer 2 device, right? Not really. All you have to do to enable a PC or Macintosh to communicate using Ethernet is to install the physical NIC card and the driver that gets the operating system familiar with the new hardware. None of that was the installation of Ethernet software. Conversely, when you remove a NIC card, you don’t have to go into an Add/Remove applet to remove Ethernet from the computer. It goes away with the NIC card. As a result, we can surmise that the NIC card supplies both the Physical layer and the Data Link layer (at least the MAC sublayer) Ethernet functionality of your PC. Figure 2.7 shows an example of a network interface card.

FIGURE 2 . 7 A sample network interface card

The Transceiver

In the strictest definition, a transceiver is the part of any network interface that transmits and receives network signals (transmitter/receiver). Every functioning network interface has a transceiver, internal or external. Those that do not have a built-in transceiver (for example, NICs with only a DIX/AUI port) will require an external transceiver, but every interface requires some

form of transceiver to convert the device’s digital signal to one that is compatible with the network medium. The appearance and function of the external transceiver vary with the type of

network cable and topology in use.

Note:

Some network interface cards have an Attachment Unit Interface (AUI) port (typically a 15-pin D-shell [DB-15] connector), with no internal transceiver, that allows an external transceiver to be used, thus changing the media types to which the NIC can connect. This port is more accurately known as a DIX port because AUI was originally reserved for the drop cable that connected to the DIX port, but through common use, AUI surpassed DIX in popularity. For example, if you are using an Ethernet 10Base2 network interface card with an AUI port, you can connect to an Ethernet 10Base-T network by using an external transceiver attached to the AUI port.

The Repeater

The simplest of all the Physical layer devices is the repeater, which simply regenerates the signals it receives on one port and sends (or “repeats”) them on another as if it were the original physical

source of the transmission. Contrast this functionality to an analog amplifier, sometimes referred to inaccurately as an analog repeater. The analog device is unable to completely discern what part of the incoming signal is intentional and what part of it is noise. As a result, except for with high-end models that can incompletely mitigate the noise to some degree, all of the interference is amplified, along with the intended signal. Digital repeaters used in early networking, and still seen in various outposts today, were not subject to such noise reproduction. 

      Repeaters are used to extend the maximum length of a network segment. They are often used

if a few network devices are located far from the rest of the network. Figure 2.8 shows a network that uses a repeater.

There is a limit to the number of 10Mbps repeaters that can be used in serial on a network without separating them by at least a layer 2 device. The 5-4-3 Rule dictates how many repeaters can be used on a network and where they can be placed. According to this rule, a

single network can have five network segments connected by four repeaters, with three of the segments populated. The other two segments are simply for inter-repeater connectivity. The

5-4-3 Rule ensured that the minimum-sized Ethernet frame of 64 bytes could begin being received by the destination device before the last bit was transmitted by the source device. If this rule is violated, two devices may not be able to reach one another across the network. Furthermore, a phenomenon known as late collisions becomes more prominent, resulting in improper recovery behavior by the transmitter, which already believes the frame has made it

across the network safely and does not hear the collision. Errored frames known as runts are often the product of late collisions. Today’s faster hubs are not bound by the 5-4-3 Rule but

actually by stricter guidelines because the data appears compressed by 10Mbps standards and cannot tolerate as long of an electrical distance between transmitting and receiving devices.

Figure 2.9 illustrates the 5-4-3 Rule.

FIGURE 2 . 8 A repeater installed on a network

FIGURE 2 . 9 The 5-4-3 Rule for network repeaters

FIGURE 2 . 1 0 A standard hub

The Hub

After the NIC, a hub is probably the most common Physical layer device found on networks today. A hub (also called an Ethernet concentrator) serves as a central connection point for several

network devices. At its basic level, an active hub is nothing more than a multiport repeater. A hub repeats what it receives on one port to all other ports, including the port on which the signal was received, so that the transmitting device may monitor and recover from collisions. 10Mbps hubs are, therefore, also subject to the 5-4-3 Rule.

There are many classifications of hubs, but two of the most important are active and passive:   

    An active hub is usually powered and it actually regenerates and cleans up the signal it receives, thus doubling the effective segment distance limitation for the specific topology (for example, extending a twisted-pair Ethernet segment another 100 meters).

     A passive hub is typically unpowered and makes only physical, electrical connections. Typically, the maximum segment distance of a particular topology is shortened because the hub takes some power away from the signal strength in order to do its job. You should not expect to see these in service anymore.

The Multistation Access Unit (MAU)

The Multistation Access Unit (MAU) is a Physical layer device that is unique to Token Ring networks. Token Ring networks use a physical star topology, yet they use a logical ring topology.

Logical topologies are discussed in the upcoming section “Data Link Layer.” The central device on an Ethernet star topology network is a hub, but on a Token Ring network, the central device

is a MAU (sometimes called MSAU, for those who prefer to represent the word station separately in the acronym).

The functionality of the MAU is similar to that of a hub in that active MAUs regenerate the signal they receive as they send it out, but the MAU provides the data path that creates the logical

“ring” in a Token Ring network. Unlike a hub, the MAU passes the bits received on one port to the port that the MAU deems the nearest active downstream port. In doing so, the MAU recognizes which ports have active stations attached and bypasses any inactive ports in its search for the next active downstream port. The data can travel in an endless loop between stations. MAUs are chained together by connecting the Ring Out port of one MAU to the Ring In port of another and connecting the last Ring Out port to the Ring In of the first MAU in the chain, thus forming a complete loop. MAUs on the market since the mid ‘90s were found fairly

reliably to have a feature that allowed the ring to be completed internally, without the last MAU connecting back to the first. Such flexibility resulted in considerably more expansive rings, without

the restriction of that potentially longer run back to the beginning if expansion had occurred in a straight line. In a Token Ring network, you can have up to 33 MAUs chained together. MAUs are shown in Figure 2.11.

FIGURE 2 . 1 1 MAUs in a Token Ring network