Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I am comparing my cable TV wire to my network wire. In a TV cable wire, different frequencies are used by different TV channels. And since a certain channel use a fixed frequency, I think the only left way to represent different signal is with the carrier wave's amplitude.

But what about the network wire? For all the network cards with the same type, do they also use different frequencies to send signals just like TV cable? I vaguely remember that they use frequency adjustment to represent signals. So the frequency should not be a fixed one. So how did all the network cards that sharing the same medium differentiate their own signal from others?

share|improve this question

migrated from stackoverflow.com Jan 22 '11 at 10:21

This question came from our site for professional and enthusiast programmers.

    
Thanks studiohack, your tags are much better. –  smwikipedia Jan 26 '11 at 13:45
    
Your statement "I think the only left way to represent different signal is with the carrier wave's amplitude" is incorrect. There are lots of different ways to modulate a carrier to convey information: amplitude modulation (AM) as you say, frequency modulation (FM), phase modulation (PM), etc. en.wikipedia.org/wiki/Modulation All of these, including AM, necessarily introduce power at frequencies other than the carrier frequency--that is, they require a certain bandwidth of frequencies. You may not have appreciated that that is true of AM. –  coneslayer Jan 28 '11 at 14:42

6 Answers 6

Assuming you're asking about how the Ethernet PHY layer handles multiplexing, here's a simplified answer.

Each network consists of at least 2 stations sharing the same medium (wire). More than 2 stations sharing the wire is pretty common with Ethernet. Because they're sharing a medium, only one station can "communicate" at a time. First-come-first-served is how they decide who gets to use the medium.

To prevent the network from degrading too much due to collisions (more than 1 station sending at the same time), Ethernet uses CSMA/CD (carrier sense multiple access with collision detection) which in laymen's terms means "if no one's using the wire, try to send. If you detect a collision while sending, stop and wait a random amount of time before trying again".

share|improve this answer
1  
So, you mean it is actually just a time division multiplexing. Thus only one frequency is sufficient. It seems my knowledge of frequency division multiplexing make me think too much. The solution turns out so direct and simple. Please correct me if I am wrong. Many thanks. –  smwikipedia Jan 22 '11 at 4:45
1  
@smwikipedia: Different multiplexing techniques can be applied, see the Multiplexing Techniques box at right side of this page –  Tom Wijsman Jan 24 '11 at 2:12
    
Yes, Ethernet does a time multiplexing, it does not allow two simultaneous senders on the wire ('ether'). –  9000 Jan 24 '11 at 2:12

They don't. The differentiation is done in higher up level of the OSI model. (Your TCP/IP stack) takes care of addressing... Simple answer -> Each data gram (or chunk of binary data that comes out/in to your nic) has 2 MAC addresses and 2 IP address. One pair for the sender and one for the recipient. (I am a bit rusty on this as i studied it over 12 years ago)...

In any case when computer a sends something to b and they are both on the same wire, the addresses contained within the datagram differentiate who the recipient and senders are. The hardware just passes the info along, that's all.

share|improve this answer

Frequency on the network has to do with to the number of cycles per second, and relates more to the speed or bandwidth of the transmission.

Ethernet over twisted pair has no frequency-division multiplexing (FDM) or other frequency shifting modulation in use; each signal has full control of wire, on a single frequency. The transmission of data is carried out by voltage changes. A packet rolling down the Ethernet cable is represented as a series of voltage changes, each with a corresponding change in current. When a signal is impressed on a line (when a host sends a packet on the Ethernet), the voltage at the end of the wire must go from 0 to -2.5 volts.

Otherwise, devices from different manufacturers could not talk together and adapt their speeds/frequency to a common value if they also had to agree on a common frequency.

share|improve this answer
    
In addition, there is not time division multiplexing. Both FDM and TDM need to be coordinated by some central agency. Ethernet operates by performing collision detection/avoidance, much like @huio says. –  Rolnik Jan 26 '11 at 0:13

you gave little detail, so i guess you are talking about multiplexing. is that close?

share|improve this answer
    
yeah, it is multiplexing that makes think of this question. –  smwikipedia Jan 9 '11 at 6:42
    
could you define 'medium' shared by network cards (card-router wire, card-router-router-router-...-backbone wires)? –  What is the Question Jan 9 '11 at 18:33
    
@(What is the Question), Your reply give me the implication that different medium use different multiplexing techonologies? My initial concern is Ethernet, but now I don't want to limit my question to Ethernet only. Any medium could be discussed. Are they different in multiplexing? –  smwikipedia Jan 26 '11 at 13:49

First of all you have to know what multiplexing means. Multiplexing is transferring different date at the same time over the same wire.

TV cables are "broadband" cables and implement frequency multiplexing. The frequencies don't disturb each other since their sinus waves don't touch each other.

The cables used for Ethernet (and other networking protocols) don't implemend multiplexing in that way. They just use CDMA/CD, the CDMA stands for Carrier Sense Multiple Access. This means that NICs have to acquire the "right" to send. This is solved in different ways:

The CD means "Collision detection". This means that collisions are not prevented, they are only detected. If the NICs detect a collision, they stop sending data for a random amount of time, and continue sending afterwards. The "right" to send is acquired by detecting if the line is free (

There are also other types, such as CDMA/CA. This is called Collision Avoidance, used iN WLAN. Hosts are not sending randomly but rather waiting until the line is free (thus never colliding in the first place)

share|improve this answer
    
Also, with ethernet switches, there's usually no need for collision detection: Simply speaking, the switch sends everything over its backplane (its bus), and establishes connections only between the ports communicating with each other. –  Tobias Plutat Jan 27 '11 at 13:17
    
There will still be colissions on the cable between the router and the PC. I know that collission detection was made for when hubs or bus cables were the choice of network administrators... –  sinni800 Jan 28 '11 at 7:44

Cable TV uses the same channel multiplexing as aerial broadcasting, e.g each channel is given its own frequency range or bandwidth, which usually is referred to as FDMA.

A TV signal does not contain one single frequency, actually a fixed unmodulated carrier occupies 0 Hz bandwidth. It is when it is modulated it occupies bandwidth proportial to the modulation rate (symbol rate or baudrate) and the modulation shape/waveform.

A digital bitstream of ones and zeros generates a wideband signal which is not suitable for use in an environment where each communication channel has to be constrained within a given frequency band. Digital communication systems utilises more frequency conservative modulation schemes such as QAM or OFDM/DMT (the later is actually multiple QAM signals in parallell).

But in a twisted pair network cable, only one transmitter/receiver pair utilises each channel and full duplex is achieved by using one pair in each direction. So there is no major restriction on the bandwidth used by the modulation scheme, other than the bandwidth of the cable itself. Up to 100 Mbit/s is quite straightforward but Gigabit Ethernet requires more advanced modulation since the TP cable distorts the signal too much if a raw wideband (several 100 MHz) signal is transmitted. Digital uncompressed HDTV (1.5 GBit) used to be transmitted directly over coaxial cable for a shorter distance of up to 100 ft, but that would not work on TP cables. HDMI transmits the raw bitstream on TP over a few metres.

Apart from that, TP cables suffer from crosstalk and that is a limiting factor in high speed communication on the telephone local loop used by ADSL/VDSL2.

So an ethernet network card for twisted pair do not use a "tuner" to receive the signal such as the tuner in a TV. It uses a receiver that can "lock" to the baudrate/symbol rate of the transmitter. That is why the Ethernet frame consists of a preamble with a repeated pattern that the receiver can use to lock and adjust its receiver clock on (And also estimate and compensate for channel distortion introduced when using long cables).

If several network cards share the same medium, such as with coax networks with mutliple stations or TP cables and a hub, then time division multiple access is utilised using various schemes as described by the other answers.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.