Quadrature amplitude modulation, or QAM, represents symbols by varying both in-phase and quadrature amplitudes.

A rectangular $M$-QAM constellation places multiple allowed amplitude levels on each axis. For example, 16-QAM has

$$M=16$$

points and therefore carries

$$\log_2 16=4$$

bits per symbol.

Compared with QPSK, higher-order QAM can transmit more bits at the same symbol rate. The cost is denser constellation packing: for a fixed constellation scale, neighboring points lie closer together and are more easily confused by disturbances.

The constellation can also be shifted, scaled or rotated by the channel. Practical receivers therefore need sufficiently accurate gain, phase and timing recovery before applying decision boundaries.

QAM illustrates a central trade-off: increasing bits per symbol can raise data rate without increasing symbol rate, but leaves less separation between alternatives and therefore demands more accurate reception.