Volume Number: 16 (2000)
Issue Number: 7
Column Tag: Getting Started
by John C. Welch
Last time around, we covered the sub-OSI parts of a network model, AKA Layer 0. We went over the physical wiring, and wiring types that networks use. We also covered connections between stations on a network, the pinouts of Ethernet cabling, and some basic information on fiber optic connections. Finally, we went over the various network topologies. This month we move up into the actual OSI model, starting with Layer 1, the Physical Layer.
The Physical Layer is the foundation for all the other layers in the OSI stack. It contains all of the specifications for any functions that deal with the transmission and reception of signals. There are four basic functions of the Physical Layer: Mechanical, Electrical, Functional, and Procedural. These cover all the functions that are needed for the transfer of electrical and / or optical signals and data. Examples of these are signaling methods, current requirements, components on the Network Interface Card, (NIC), and the characteristics of the connector on the NIC.
One of the common problems with talking about the Physical Layer is remembering its scope. In general, if it's outside of the host, it's outside of the Physical Layer, and the OSI model in general. This includes cabling, switches, routers and hubs. The usual boundary for the Physical Layer is the connector on the NIC. Part of the reason for the confusion over the scope of the layer is that it defines a lot of what the media and other Layer 0 components have to do, but doesn't actually include those components.
Since the Physical Layer transmits and receives data, let's look at what it has to do for those functions. For transmission, the Physical Layer has to:
One important fact to note is that the Physical Layer does no error correction. It has no ability to detect errors in the data stream or frames it deals with. All it does is transfer data.
Signal Encoding
The Physical Layer primarily deals with getting signals in and out of the host, and encoding and decoding them. Regardless of the type of media used, these signals are all electromagnetic waves of some sort. Every electromagnetic wave has certain characteristics, the most obvious being its period or frequency. The frequency of a wave is the number of oscillations it makes every second. No wave is a straight line, but rather is a regular, reasonably symmetric curve that most resembles a sine wave. A diagram is shown in Figure 1.
Figure 1.
The symbol for wavelength is lambda, or l. The wavelength is the physical size of the signal through one complete oscillation, or cycle. The wavelength is also the inverse of the frequency. The frequency of a signal is the number of cycles it has in a given amount of time, usually one second. The measurement used to identify frequency is hertz, or Hz. If a signal has one cycle every second, its frequency is one Hz. If it has a thousand cycles per second, then it has a frequency of one KHz, and so on. Since frequency and wavelength are inverses of each other, the value of one can give you the value of the other. For example, the frequency of many Radar guns used by the police is 34GHz, or 34 billion cycles per second. Taking the inverse of this, using the formula 1/frequency, we get 1/34 x 109, or 2.9411 x 10-11, or a wavelength of .0294118 nanometers. Since nanometers are billionths of a meter, this is a very small wavelength indeed.
One of the more useful side effects of frequency has to do with the ability of a signal to transmit data. Since most encoding methods use the number of changes in a wave, the higher the frequency, the more data that can be transmitted by that wave. The disadvantage of higher frequencies is that the signal has less range and persistence. This is why faster LAN technologies tend to have less range than slower ones. (Note: I say 'tend to' here because when you go from copper - based LANs to fiber - optic based LANs, this falls apart somewhat, as a optical LAN has more range than a copper LAN, yet operates at much higher frequencies.) If we look at the signal in figure one above, you can see that the signal is either above or below a central point. The amount of distance between the top or bottom most point in the signal, and the central point is the amplitude, or strength of the signal. Normally, amplitude is measured in positive or negative volts, so our central point is zero volts. The positive and negative sections of the signal are states, and the transition points at zero volts are the state changes. As we will see, state changes are critical to encoding data in a signal. There are a few more parts of a wave that we need to understand, so we'll take a look at another wave form, the square wave, in Figure 2.
Figure 2.
The square wave is named for its shape, and is a common wave form used in the transmission of data. The square wave is also useful for identifying certain parts of a wave that are used in signaling and encoding. The leftmost edge of the wave is the leading, or rising edge. The upper and lowermost parts of the wave are the peaks, and the middle line, where the wave is transitioning from a high to a low is the trailing, or falling edge. All parts of the wave can be used in signaling and data encoding, depending on which method is used.
Regardless of the method chosen, there are three basic ways to modulate a wave so as to allow it to carry data. (In truth, these are the basis for many, many more modulation techniques, and other trucks to get as many bits per cycle as possible onto a wave, but for our purposes, these three will do.) The first method is amplitude modulation, also known as AM or ASK, for Amplitude Shift keying. Since the amplitude of a signal is its strength, the way this works is suggested by the name. If the data being carried by the signal is a binary 1, then the pulse amplitude is increased by a given amount. If the data is a binary 0 then the pulse amplitude is decreased by the same amount. A middle amplitude indicates no data. Amplitude Modulation is simple to implement, and widely used for voice transmission, as in AM radio. Due to limitations in the modulation technique itself, ASK is only used for transmissions up to 1200 baud.
The next method used is Frequency Modulation, a.k.a. FM or FSK. This functions by introducing a second signal into the first signal. If a binary 1 needs to be transmitted, the frequency of the first signal is upshifted, or increased. If a binary 0 is being transmitted, the signal frequency is downshifted, or decreased. If no data is being transmitted, the frequency is left alone. FSK is more efficient and reliable than ASK, and is used for data rates up to 1200 baud.
The final method is known as Phase Modulation, or PSK. In this method, the phase of the signal is altered to indicate a binary 0 or 1. If a binary 0 is sent, the phase is not altered. If a binary 1 is sent, the phase is switched to its opposite. This allows one bit per shift(s) to be sent. If more than one phase shift is used, then more bits can be sent in a set of shifts, or burst. PSK is the most efficient, and reliable of the three modulation techniques, allowing data rates of up to 9600bps. These may not seem like fast data rates, but in combination with other encoding methods, (most of which we will cover in other articles, as they are protocol - dependant, and therefore beyond the scope of the Physical Layer.), we get the data rates we are accustomed to on today's LANs.
Whichever form of modulation and encoding is chosen, that is then modified by other methods to allow the signal to pack more bits per cycle, thereby increasing efficiency of the signal as a data carrier. Depending on the methods used, and the frequency of the signal, it is possible to encode anywhere from two to forty bits per cycle in a signal. So a cable that has a maximum bandwidth of 500MHz is capable of transmitting several gigabits of data per second. Since fiber optic cables can handle much higher frequencies than copper, it is obvious why fiber is becoming more popular as a way to move the staggering amounts of data that a modern network is required to move.
One thing to remember is that any form of encoding, no matter how fast or efficient, imposes an overhead that subtracts from the bits a signal can carry. As an example, Gigabit Ethernet uses an encoding scheme known as 4B/5B. 4B/5B states that for every 4 - bit string of data transmitted, a 5 - bit pattern is used to encode it. This means that 20% of the data transmitted is overhead. If Gigabit Ethernet were to be strictly defined as 1000, or even 1024Mbps, then the best speed you could actually get would be in the 800Mbps range. To get around this, Gigabit Ethernet is actually defined as 1250Mbps, so when the encoding and other overhead is accounted for, the usable speed is still very close to 1Gbps.
It may seem that we have glossed over a lot of the Physical Layer. That is partially true. Much of the Physical Layer deals with extremely low level electrical issues that would be of only academic use to the network administrators. Since we covered physical media in the previous article, there is no need to re-hash that here. Also, anything that is protocol - specific is the property of the higher layers. The important thing to remember is that the Physical Layer is only concerned with ones and zeros, and the transfer of those things. The next article will deal with the Data Link Layer , and the Medium Access Control sublayer, and that is where we will get into much of the protocol specific issues of a network.
John Welch <jwelch@aer.com> is the Mac and PC Administrator for AER Inc., a weather and atmospheric science company in Cambridge, Mass. He has over fifteen years of experience at making computers work. His specialties are figuring out ways to make the Mac do what nobody thinks it can, and showing that the Mac is the superior administrative platform.
