PSK and QAM modulation have two advantages over ASK:

*They are not as susceptible to noise.

*Each signal change can represent more than one bit
PSK Disadvantage
more complex signal detection / recovery process,
than in ASK and FSK
QAM advantage:
data rate = 2 bits per bit-interval!
higher data rate than in PSK (2 bits per bit interval), while bandwidth occupancy remains the same
4-PSK can easily be extended to 8-PSK, i.e. n-PSK
however, higher rate PSK schemes are limited by the ability of equipment to distinguish small differences in
phase uses "two-dimensional" signaling
original information stream is split into two sequences that consist of odd and even symbols
PSK modulators are often designed using the QAM principle, but are not considered as QAM since the
amplitude of the modulated carrier signal is constant. QAM is used extensively as a modulation scheme for
digital telecommunication systems. Arbitrarily high spectral efficiencies can be achieved with QAM by setting a
suitable constellation size, limited only by the noise level and linearity of the communications channel.
Noise immunity of QAM is very high.
QAM is best suitable for high bit rates.
Low error probability.
Baud rate is half the bit rate therefore more effective utilization of the available bandwidth of the transmission
channel.
How QAM is better than other mechanisms used in Digital/
Analog modulation?.
It's not really fair to compare a digital modulation to an analog one because they do very different things. The best
modulation method depends on the application
QAM, like all digital modulation schemes is much more resilient to signal degradation from noise. Digital signals can
also carry a lot of information that is machine readable, more tightly compressed than an analog signal.

Compared to other digital schemes, QAM can carry 2 or more bits per cycle (per 1 Hz of frequency spectrum),
whereas shift keying methods (PSK/ASK/FSK) typically only get a maximum of 1 bit per cycle. Compared to spread
spectrum technologies, QAM is more susceptible to interference, but it is generally more data-dense.
Source:
Fading is most important thing when you design any RF Design lets understand what is fading
and how its effect.
The communication between the base station and mobile station in mobile systems is mostly
non-LOS.
The LOS path between the transmitter and the receiver is affected by terrain and obstructed
by buildings and other objects.
The mobile station is also moving in different directions at different speeds.
The RF signal from the transmitter is scattered by reflection and diffraction and reaches the
receiver through many non-LOS paths.
This non-LOS path causes long-term and short term fluctuations in the form of log-normal
fading and rayleigh and rician fading, which degrades the performance of the RF channel.
LONG TERM FADING
Terrain configuration & man made environment causes long-term fading.
Due to various shadowing and terrain effects the signal level measured on a circle around
base station shows some random fluctuations around the mean value of received signal
strength.
The long-term fades in signal strength, r, caused by the terrain configuration and man made
environments form a log-normal distribution, i.e the mean received signal strength, r, varies
log-normally in dB if the signal strength is measured over a distance of at least 40l.
Experimentally it has been determined that the standard deviation, s, of the mean received
signal strength, r, lies between 8 to 12 dB with the higher s generally found in large urban
areas.
RAYLEIGH FADING
This phenomenon is due to multipath propagation of the signal.
The Rayleigh fading is applicable to obstructed propagation paths.
All the signals are NLOS signals and there is no dominant direct path.
Signals from all paths have comparable signal strengths.
The instantaneous received power seen by a moving antenna becomes a random variable
depending on the location of the antenna.

RICEAN FADING
This phenomenon is due to multipath propagation of the signal.
In this case there is a partially scattered field.
One dominant signal.
Others are weaker.

difference between qam and qpsk
That is correct. QPSK is a harder to implement due to the 45 degrees phase difference between
quadrature components. it is preferred to use a IQ modem So in conclusion, they are both the same
theoretically but the implementation is different. I'd prefer QAM as I said. you do not have to
implement the 45 degree phase shift in both transmitter and receiver which might degrade
performance.
ook. the quadrature components can be described as follow:

S(t) = I(t)*cos(Wc*t) - Q(t)*sin(Wc*t)

in which Wc is the carrier or IF frequency of the system. this is the output of a quadrature modulator.
Imagine the 2 dimensional signal space of which the bases are cos(Wc*t) and -sin(Wc*t). Thus, Q(t)
defines the imaginary part of the constellation point while I(t) describes the real part of the
constellation point.
when you're using a QPSK modualtor of which the constellation points are {-1,+1,+j,-j} each symbol
uses only cos(Wc*t) or sin(Wc*t). you just need an oscillator and a 90 degrees phase shifter to
introduce the quadrature component.
when you are using the constellation points 1/sqrt(2) * {+1+j, -1-j, +1-j, -1+j} each symbol has
both real and imaginary parts. you have shift it like 45 degrees compared to the latter case, to
achieve this. or you may also be able to use the second case. you have to be careful when
demodulation because each symbol has a 45 degrees phase shift compared to the quadrature
components. that's the 45 degrees I'm talking about not the constelltion points compared to each
other. while demodulation, you have keep this 45 degrees very accurate not to confront error. I hope I
have made my point. ;)
SYMBOL ERROR RATE CALCULATION FOR
PAM AND QAM
The general equation to calculate symbol error rate (Pe) in PAM and QAM systems is [1]:
PeKQ(3SNR21)(1)

where K is the average number of nearest neighbor points on the constellation, Q(.)is related to the well
known complementary error function as:

Q(x)=12erfc(x2)(2)
and is the spectral efficiency of the signaling method and is defined as:

=log2MBT(3)
where B is the channel bandwidth and T is the signaling symbol period, and M is the number of points on the
constellation.
For baseband systems the maximum value for the symbol rate 1/T is two times of the channel bandwidth, so
the maximum value for spectral efficiency is when: BT=2. A passband channel with a bandwidth of B is
equivalent to a complex baseband channel with bandwidth of B/2, so the maximum symbol rate will be 1T=B,
or we have: BT=1.
Here are examples of PAM2 (BPSK) and QAM16 constellations:

For PAM2 the number of nearest neighbors is 1 and for QAM16 it is 4 for all the points in the center, and 3 for
points along the side of the square except points at four corners of the square, which it is 2 for them. So the
average number of nearest neighbors for QAM16 is 3. For larger PAM constellations K approaches to 2, and
for larger size QAM constellations K approaches to 4.
If we define the rate-normalized SNR (SNRnorm) as:
SNRnorm=SNR(21)(4)
Then we will have the generalized relation between the probability of symbol error and SNR
independent of constellation as:
PeKQ(3SNRnorm)(5)
If we plot this equation for a range of K values from 1 to 4 we get the following figure:

As it can be seen there is a small difference between the probability of symbol error for different values of K.
So this curve is extremely useful to quickly calculate the required SNR for a given probability of error, or the
probability of error for a given SNR.
For example if we want to design a QAM system with 256 points on the constellation and the required symbol
error rate is 1010, by looking at the above figure one can see that we need a normalized SNR of about 11.5
dB. To see what would be the actual required SNR, we then can use [3] which gives us =8 and then use [4] to
get to the actual SNR as SNR=(281)SNRnorm which is about 35.6 dB.