[Back to Lecture Notes page]
Sub-topic Outline:
This category of media can also be called "guided" media, since signals are bounded and guided by physical boundaries (eg a cable).
Eg. floppy disks, CD's, magnetic tapes.
Can have high bandwidth, but also high delay.
It can be high bandwidth because we can put a HUGE amount onto disks on a truck, but obviously it wouldn't be as fast as transmitting through a cable or through statellites, etc.
Usually the most cost effective when sending large volumes.
In most cases of sending large volume of data, if time is extremely urgent, it is cheaper to send normal post.
Used by telephone system
I refer to the telephone lines coming out of homes and offices. They are still based on twisted pair. A lot of the major line connecting up main switching stations uses more advanced media these days, like fiber optics. More on these media, and the telephone system, later.
A lot of twisted pair wires run together side-by-side, such as from all the telephones from homes in a single building. If not for the twists in all the pairs, the interference would make them unusable.
Twisted pair wire usually only run for a few kilometers.
Different categories of twisted pair wires, two important ones for computer networking:
Used mainly for phone lines before 1988. Most building has a Category 3 cable running from a central "wiring closet" to each floor into each office.
Both these type of wiring is also called Unshielded Twisted Pair (UTP). The word "unshielded" comes from comparing with the heavily shielded wires produced by IBM in early 1980's.
Common name "coax". Baseband coax are constructed for transmitting in digital. Coax designed for analog transmission are called Bradband coax (see next section). Baseband coax usually run for a few kms - anything further required an amplifier.
Fig 2-3 Tanenbaum textbook p 84.
Baseband Coax used to be widely used in connecting switching stations in the telephone system, but they are replaced with fiber optics now.
Cable TV is the main remaining user of coaxial cables now, since the adoption of fiber optics for a lot of other purposes.
Fig 2-4 Tananebaum textbook p86
Dual cable systems have two cables running in parallel at all parts of the network. The network is in the form of a tree. One cable is for sending data to the root (called "head-end") of the tree, and the other is for the head-end to broadcast to every machine in the tree.
Single cable systems only have one cable for both sending to the head-end, and for broadcasting back to the machines. To do this, there must be a way of identifying whether a signal detected is meant to be for the head-end, or is actually from the head-end to the particular machine. This is done by allocating a band of frequency inbound signals (FROM the head-end) and another band for outbound signals (TO the head-end). In a subsplitsystem, we use 5 - 30 MHz for inbound signals, and 40 - 300 Mhz for outbound. In a midsplitsystem, we use 5 - 116 MHz for inbound signals, and 168 - 300 Mhz for outbound.
Technically, broadband coax is inferior to baseband in transmitting digital signal, but it is used regularly for digital communications because so many broadband coax networks are available.
The presence of light means 1 bit. No light means 0 bit.
3 components:
Fig 2-5 Tanenbaum textbook p88
The principal of fiber optics is based on the fact that when light is travelling within glass, and it hits the boundary of the glass, if the angle it hits the glass is more than a certain angle, it will be reflected back and remain in the glass (until it reaches the boundary on the other side. If we can keep the angle which the ray of light hits the boundary to always be more than a certain angle, it will always remain in the glass as it travels. See figure 2-5(b).
The multiple rays of light are all reflecting off the boundary of the glass, travelling through it without interfering with each other.
If we reduce the diameter of the glass to a certain extent, we can concentrate teh direction of the light beam so that it travels in a perfectly straight line without touching the boundary. These single mode fibers can transmit the light to longer distances.
Attenuation of light of a given distance
The reason we want to know what attenuation light of a certain frequency has, is because we want to select the best frequencies to use as our light source for fiber optic transmission. We want to select the range of frequencies where attenuation is minimal. Figure 2-6 shows 3 band of frequencies where the attenuation is lowest, so we should use those frequencies as our light in the fibers. Note that although the band on the left (centred at 0.85 microns) is not actually low relative to the rest of the graph, using the frequencies in that band allows the lasers and the electronics to be made from the same material.
Fig 2-6 Tanenbaum textbook p89
Fig 2-7 Tanenbaum textbook p90
Very thin glass core:
Fiber optic cables on ground are laid within a meter of the surface. On the shore, they are burried in trenches. Under the sea, the cables just lay on the surface of the seabed.
3 ways of connecting the ends of fibers to form a longer line:
Having connectors also allows us to plug the cables into sockets.
Special sleeves exists where we just align the ends of the fibers (the ends must be carefully cut) side by side and clamped in place.
Two fibers melted together.
2 ways of emitting light signal (Fig 2-8 textbook p91):
Fig 2-9 Tanenbaum textbook p92
This is one example configuration of a fiber optic network. Here we have a central ring of fiber optic cable, where light signal is travelling. Each computer hooks up to the main fiber using an interface. The interface receives a signal, converts it to electrical signal, and sends it to the computer. It also regenerates the light at full strength and retransmit the light signal. This is called and active repeater, because it actively regenerates and retransmits the signal.
We can also have passive interfaces for the computer. In this case, each computer has two taps to the central ring. One is a phtodiode to detect the light, and another is a light generator to transmit if the computer wants to transmit a signal. The taps are called passive because they only detect the light and do not intercept it. So unlike active repeater, if the interface for a computer goes down, the network still runs. It just means the computer using that interface will not be ont he network. This is different from the previous example, where if an active repeater goes down, the whole network goes down.
Fig 2-10 Tanenbaum textbook p93
In this example, the configuration is a star. Again, the main hub (the cylinder in the figure) can be active or passive. Active hubs receives signals using photodiodes and retransmit them with maximal strength to reach all receivers. Passive hubs only distributes the incoming light signal to all outgoing lines. In the case of passive hubs, the transmitter from the originating computers have to ensure they transmit a strong enough signal to be able to get to all receivers once the light is distributed.
normal fiber optic lines can go up to 30km without needing a repeater, but copper wires can only go up to about 5km.
..compared to tapping electrical signals on a wire.
There must be one light generator on one side, and one photodiode receiver on the other side. We can make use of the same fiber for two sets of generator/receiver pair, both transmitting light in opposite directions, but it becomes very hard to synchronise the rays of light.
..and also the cost for fiber interfaces as well (like connectors, photodiodes, lasers, etc).
What do you expect from a glass core?
Fiber optics is almost always the prefered medium of use for guided transmission.
[Back to Lecture Notes page]