Re: raised cosine filter
A rectangular pulse has very high bandwidth (infinte in thoery), due to the 0-time sign changes in the waveform. If we use the pulse to straightaway modulate a carrier sinewave, the changes in the sinewave will be very abrupt ie; the bandwidth of the carrier will also be very high. Most channels have limited bandwidth and so, they will heavily attenuate/distort the carrier wave. Thus it is very desirable to make the carrier occupy the least possible bandwidth for any given data rate. In other words, it would require a channel of infinite bandwidth to carry such a waveform without distortion.
Naturally reduction in distortion can be achieved when the modulating signal occupies less bandwidth. This would mean that the pulses would have to get longer and longer However, we dont want to reduce the data rate either. So the question comes down to this: I want to transmit at x symbols per second, what is the minimum bandwidth that I can occupy? It turns out to be half the symbol rate (x/2 Hz). (Nyquist theorem). So the next question is which waveform will give me that bandwidth for transmitting x bps? Since we need, in this case, striclty limited BW, the pulse has to be infinite duration and so is not realisable. The raised cosine wave makes the practical closest-cut to this reqiurement. Also by virtue of its zero value at the peak of the adjacent symbols, so that also inter symbol interferene is kept to a minimum.
-b