Capacity (a data rate) is never defined as an SNR or Eb/No, and capacity *is* an upper limit, *below* which it is possible to have essentially error free transmission.
By definition capacity, measured in binary digits, is
P+N W log2 ----- N
bits per second at some arbitrarily small error rate, which some sufficiently complex encoding system would be capable of accomplishing.
By definition it is not possible to transmit at a higher rate, regardless of the efficiency of the encoding system, without a "positive frequency of errors".
System capacity, not channel capacity. That is to say that the channel has a capacity determined by the bandwidth and the SNR; but any given system may not be sufficiently involved to attain that capacity, or to make use of either the bandwidth or the SNR. Hence the system has a capacity that is something less than the channel capacity. We could refer to the ratio of the two as "system efficiency" if you like.
Consider a typical POTS line using twisted pair. If we ignore other cable pairs, the channel it provides has limited bandwidth and unlimited (relatively) SNR. Which is to say that no matter what we do, it will not transmit a 10 Mhz signal very far, but if we need 1 dB more SNR all that is required is to feed a signal with 1 dB additional amplitude.
Said line is not limited by the Shannon-Hartley theorem of channel capacity because no practical system is going to use the actual maximum power that could be applied.
Therefore the limit for system capacity would be in terms of how well the modulation/encoding scheme uses the available bandwidth. SNR is not part of the equation.
Now consider exactly the opposite type of channel, where bandwidth is not limited but SNR is. Fiber optic cable is an example (and for practical purposes satellites, up to the maximum bandwidth of the transponder, are too). There is far more bandwidth available on a typical fiber than is required for any given application. Practical systems do not use anything like the available bandwidth. The actual bandwidth used depends on the data rate and the modulation scheme. The BER depends on the SNR, not on the available channel bandwidth.
Previously I posted a chart showing SNR values for several types of digital modulation. Posted responses related to "what bandwidth" type questions, which are not relevant. The bandwidth varies with the bit rate, is not limited by the channel, and does not affect the bit error rate.
As I mentioned, I took the figures from various graphs, as that is the way it is always presented and unfortunately we cannot easily post graphs in an ASCII text message. I poked around looking for a really good example graph, and think this one expresses it better than others (if for no other reason than showing Shannon limits on the graph),
See Figure 1.
So which is it? FEC is more efficient and increases efficiency bringing the system closer to maximum capacity, or uses more bits causing the system to operate at a "reduced *information* rate".
You can't have it both ways.
(One problem is that you use "channel capacity" to describe two different things, alternating between Shannon's maximum capacity for a channel and the actual information rate of a given system.)
But *you* are telling us that he was wrong! He said it reduced the information bit rate available, and you are saying it is necessary in order for the information rate to approach channel capacity! You can't both be right...
Whatever arbitrary target values a system is designed to meet.
The bullshit meter just slammed against the peg, and bent one more time. *NOBODY* defines it in those terms. Shannon's maximum channel capacity is *defined* as a low BER, less than some arbitrary value.
See Theorems 9, 10, 11, and 20 in "A Mathematical Theory of Communication". For example,
Theorem 17: The capacity of a channel of band W perturbed by white thermal noise power N when the average transmitter power is limited to P is given by:
P+N C = W log -----. N
This means that by sufficiently involved encoding systems we can transmit binary digits at the rate P+N W log2 ----- bits per second, with arbitrarily small N frequency of errors. It is not possible to transmit at a higher rate by any encoding system without a definite positive frequency of errors.
At the expense of bandwidth (which is a channel resource that by your reasoning would be "wasted" when FEC is used).
The reason it is used *is* to reduce the BER under less than ideal circumstances. For satellite systems there is indeed the valued advantage of power efficiency, and but under normal circumstances the minimum required BER could be obtained with FEC disabled. That would *not* require increased power, and
*would* release the extra bandwidth. The circuit would function quit normally most of the time. Unfortunately, for a sufficiently significant percentage of time it would become unreliable due to high BER caused by low SNR. Correcting that with added power utilization is not as cost effective as correcting it with added bandwidth utilization.Sometimes... is about 95+% of the time. Which is not good enough if the specs call for 99.97% reliability.
It is *not* more efficient. It trades bandwidth for SNR. You pay with one or the other. It just happens that in most (not all though) instance the cost of bandwidth will be less on a satellite than the cost of power.
That statement is just as true if you reverse it and say that if you have FEC turned *ON* you are wasting channel resources.
On uses more bandwidth, off requires more power...
Get any decent book on communications link design and read it.
Here is one example:
bit rate 64 kb/ps to 44.736 Mbp/s
FEC encoding Rate 3/4 convolutonal encoding/Viterbi decoding
Modulation Four-phase Coherent PSK
Eb/No at BER (Rate 3/4 FEC) 10^-2 10^-7 10^-8 a. modems back to back 5.3 dB 8.3 dB 8.8 dB b. Through Satellite 5.7 dB 8.7 dB 9.2 dB
C/N (BER=10^-7) 9.7 dB
Nominal BER at operating point 1 x 10 ^ -7
Threshold BER 1 x 10 ^ -3
(Intelsat standards...)
Sure. See the massive confusion above...
All of the above, and a couple others too. "Make it work" was always at the top of my job description, and "it" was never well defined.
Whatever, I deleted the rest of your silly trivia and stupid insults.