Is the noiseBandwidth in the BER formulas correct?

[boulis]

Sergio claims that the FSK and PSK formulas in Castalia are wrong, we should reverse the fraction (noiseBandwidth/datarate) in the formulas.
For example, Sergio writes for FSK:

case FSK:
	return 0.5 * exp(((-0.5) * RXmode->noiseBandwidth / RXmode->datarate) * pow(10.0, (SNR_dB / 10.0)));

But it should be:

case FSK:
	return 0.5 * exp(((-0.5) * RXmode->datarate / RXmode->noiseBandwidth) * pow(10.0, (SNR_dB / 10.0)));

This claim seems counter intuitive, but we should investigate.

Read the full thread:
https://groups.google.com/forum/#!searchin/castalia-simulator/Eb$2FN0%7Csort:relevance/castalia-simulator/k8qtCjr2fk0/as62-SOIKQAJ

Some relevant material:
http://www.divilabs.com/2015/04/ebno-snr-vs-ber-curve-plotting-for-bpsk.html