| Each letter in the word is represented by | | | | noise is purely random. This theorem may be |
| different signal amplitude using thirty two scale | | | | used as a very good approximation for the |
| order. If we do so information flow speed gets | | | | ultimate channel capacity of most of the |
| greatly increased as per Hartley law. Since now | | | | transmission channels in spite of the fact that the |
| each letter is represented by one symbol instead | | | | noise present in most channels is never perfectly |
| of five. But in such a case noise will cause | | | | random. It is found that the limiting channel speed |
| numerous error making the system useless | | | | for a typical telephone channel is about thirty |
| extremely large transmitting power is used. This | | | | three kilo bits per second. How ever speeds used |
| fact may be established on comparing the power | | | | in practice over such channels do not normally |
| requirements for the binary coding system and | | | | exceed eleven kilo bits per second. Doubling the |
| any N-level system under the same noise | | | | speed the bandwidth of a noise limited channel will |
| conditions. For a given transmission and coding | | | | double its capacity would amount to |
| system there results are threshold noise levels | | | | misinterpretation. Actually the capacity gets |
| below which practically no errors occur due to | | | | increased by only eighty percent depending upon |
| noise. Using binary code noise has to compete | | | | signal to noise ratio. Thus we see that there |
| with the full power of the transmitter to cause | | | | exists possibility of trading bandwidth for signal to |
| any serious error. It is found that in a practical | | | | noise ratio. It may also be noted that a low |
| channel, signal to noise ratio of thirty decibel | | | | channel capacity does not mean that the desired |
| ensures almost error free reception. This thirty | | | | amount of information can not be seen over a |
| decibel ratio implies that noise power be one by | | | | given channel. It simply means sending this |
| thousand of signal power or root mean square | | | | amount of information takes longer time. Lastly it |
| noise voltage be one by thirty one of root mean | | | | may be seen that the Shannon-Hartley theorem |
| square signal voltage. | | | | represents a fundamental limitation. Any attempt |
| The transmitted power is required to rise | | | | to exceed the Shannon limit would result in |
| tremendously if a desired high signal to noise | | | | unacceptable error rate. In good quality |
| power ratio is to be maintained on increasing | | | | transmission system maximum acceptable error |
| signaling speed that is on increasing the number of | | | | ratio is one in 1000000. All the messages sent |
| coding levels. Shannon-Hartley theorem gives the | | | | though the noise limited channel are unpredictable |
| maximum signaling speed in a channel in which the | | | | or random. |