Which method provides better noise immunity?
I know you can transmit more data via psk, but what about which method is less susceptible to noise?
Which method provides better noise immunity?
I know you can transmit more data via psk, but what about which method is less susceptible to noise?
As you say in your comment, PSK signals are more bandwidth efficient, but FSK signals have good noise rejection for their data rates. It is thus difficult to compare them on even terms.
The traditional way of measuring noise immunity is measuring/calculating the probability of a bit error given the bit energy vs. noise energy (Eb/No). This is a good way of making everything fair
The following plot uses this paper for the FSK analysis, and this page for the BPSK analysis.

As you can see, BPSK wins quite handily. I won't attempt to prove it, but QPSK has the exact same BER curve as BPSK. You can intuitively see that if you think about how the real and imaginary portions of the symbol are orthogonal to each other, so you are essentially just sending two orthogonal BPSK signals at the same time. The BER curves of the higher PSK types (e.g. 8-PSK), though, do get worse because the symbols are closer together, so more energy is needed to distinguish them in the presence of noise.
So why does anyone use FSK? Mostly because it's pretty simple to implement, particularly the non-coherent receivers.