Megahertz to Megabits Conversion?

[Nortelstud]

Sir/Madame,

I have a question. I have been reading allot about KHZ, MHZ, Gigahertz and the electromagnetic spectrum in general. When someone mentions to me I have a link that can travel at say 8 megabits per second. I know what this figure represents as far a raw data in mbps or kbps.

Even if the link speed is 0c-12, 0c-48, 0c-192 etc I can think how fast this is and how much through-put is occurring.

I am starting to work at the lower layers and now I am talking about megahertz. With respect to a closed converged RF network. When someone says to me I have an 850 MHZ system, or if I am trying to visualize how many mpbs it allocated to a certain frequency, I cannot visualize in my head of what this actually means. How much bandwidth do I have here?

An example would be and correct me if I?m wrong. If I say I have a 6mhz channel width. I would think hey I can transmit 30 mbps in this space. And the only reason I know this is from other people saying it. Is there a "simple" Equation to calculate this?

Is there some chart I can reference to give myself an idea on how to talk between these two? One measuring the physical layer speed or bandwith of the wire and the other measuring layer 3 speeds? I have looked up and down to find this information. But can?t seem to find it.

Also Is it correct that no matter If we are transmitting across the Air, cable, or any other media inlcluding Fiber which uses light, whether it be digital or analog information we are still infact using radio waves which are part of the electromagnetic spectrum?

Thanks,

Justin

 

[Varun]

There are equations that allow you to calculate the capacity in bits per second of a transmission medium. Really the only limiting factor is noise. If you have no noise, you can transmit infinite amounts of data over any bandwidth. Of course, there is always noise so you are limited...

Anyways, there are two equations that are used.

The first is Nyquist's. C=2B*(log base2(M)) where C= channel capacity in bps, B=bandwidth of transmission medium, and M=the number of levels used in the transmission (ie the simplest is a two level system, where low=0 and high=1, but you can have multilevel systems where each level equates to multiple bits of data)

Nyquist doesn't take noise into account though. Shannon's formula is similar, but it includes noise:
C=B(log base2(1+SNR)) where once again C=capacity in bps, B=bandwidth, and SNR= Signal to noise ratio (NOT in dB)

The problem with Shannon's is it gives only a maximum theoretical capacity, and really gives no solution to achieving that capacity.

Anyways, there are many types of transmission schemes, and they all require different amounts of bandwidth. Some, like the simple NRZ-L, and some much more complex like Differential Manchester. All of the schemes take a different amount of bandwidth.

The simplest NRZ-L for instance, maxes out at 2 bits per hertz. Therefore your 6MHz transmission medium would cap out at 12Mbps. A really nice encoding scheme is MLT-3, where the worst case scenario is 4 1's which can still be done in one Hz, therefore your bandwidth would be 24Mbps.

The web has some nice articles on some of these schemes if you are more interested. Google MLT-3 and you'll be sure to find some nice articles.

There are other ways to add more levels, and therefore increase the bandwidth as well. Phase, amplitude, frequency, and combinations of all of them can be used to squeeze more transfer rate out of less bandwidth.

As far as electromagnetic waves, fiber optics do not transfer electromagnetic waves, which is their real benefit since they are not affected by electromagnetic noise. Light is a beast all its own.

 

[Born2bwire]

Originally posted by: Varun
There are equations that allow you to calculate the capacity in bits per second of a transmission medium. Really the only limiting factor is noise. If you have no noise, you can transmit infinite amounts of data over any bandwidth. Of course, there is always noise so you are limited...
As far as electromagnetic waves, fiber optics do not transfer electromagnetic waves, which is their real benefit since they are not affected by electromagnetic noise. Light is a beast all its own.

Obligatory "Light is an electromagnetic wave" post.

 

[Varun]

Light behaves like a wave sometimes I guess. It's not the same as a radio wave or sending current down a wire though.

 

[Born2bwire]

Originally posted by: Varun
Light behaves like a wave sometimes I guess. It's not the same as a radio wave or sending current down a wire though.

Yes it is. Light is an electromagnetic wave in the terahertz region. The physics that govern an electromagnetic wave through a waveguide for microwaves and radio waves are the same as that of light. The only thing differentiating a radio wave and a light wave are their frequency. A fiber optic cable is really just a dielectric waveguide.

 

[RaynorWolfcastle]

Varun is just about right (apart from the fact that light IS an EM wave, but that's a bit of a technicality). What you are talking about is, in some sense, spectral efficiency.

The maximum transmission rate is really dependent on what is the limiting factor. In RF systems, it's usually noise and/or interference, so Shannon is a good estimate of the best you can theoretically do since you are in a noise-limited channel. If you're interested, look up "turbo codes". Turbo codes were developped in the '90s and come close to the Shannon limit in an AWGN channel. Now, turbo codes are themselves only part of they equation since they only describe the channel coding scheme you are using. Once you've decided your channel coding scheme, you then have to pick another coding and modulation scheme to actually send this over your analog channel.

Taking UMTS (one of the 3G mobile telephony standards) you would take your packets/data stream coming down through the OSI stack (I can't recall off the top of my head but I think it would be coming in from the link layer). This signal is channel coded using some turbo code, the resulting data then goes through a W-CDMA encoder which is then modulated using QPSK and sent over the airwaves. Note that your actual usable data rate is diminished by the fact that you're using channel coding which includes parity checks and error correction.

There are several types of channel coding, and several modulation schemes. Basically, your channel conditions and application requirements dictate how exactly you send your data. Another example, digital television transmission over coaxial cable can use COFDM (Coded Orthogonal Frequency Division Multiplexing) which then uses some sort of QAM or QPSK scheme.

There's no hard and fast rule about how efficient you can be with a slice of spectrum; again, your channel and application dictates what you can and can't do. Your conditions over coax are very different from your conditions when using a wireless channel and are very different from when transmitting over fiber. I imagine your 6 MHz figure comes from some digital TV number, since I know that those channels are 6 MHz each; but that really doesn't apply everywhere.

Just in closing, the conditions when using fiber are usually quite different from other media. You can be limited by any number of factors in a fiber, though dispersion and non-linear mixing effects would probably be the most important. The nice thing about fiber is that you can generally assume that you have little outside interference (which is a huge concern in wireless transmissions), and that the various noise sources in your channel can be modeled in a fairly straightforward fashion.

 

[spidey07]

Well if you're speaking communications there isn't a direct coorelation.

You have bandwidth (measured in megahertz)
You have data rate (how fast a bit is clocked, measured in megabits)
Then you have throughput (actual payload throughput, measured in megabits or bytes)

For example..

100 Base-T ethernet actually runs as 125 Megabaud. But the data rate is 100 Megabits/sec. This is due to the encoding used to place the bit on the wire. Same with 802.11 wireless technologies - the data rate may indeed be 54 megabits/sec, but your throughput is MUCH less due to framing and transmission at layer1.

So in the end there isn't a direct coorelation because different technologies use different framing/encoding at layer 1 of the OSI model.