For M-QAM modulation, where e.g. $M = 64$ there are 8 amplitude levels. Each level is represented by a symbol that is further represented by bits through the relation: $$b=\log_2(M)$$ This means that each symbol is encoded by $b$ bits.
- I don't quite understand where 64 comes. Does it mean that the imaginary and real component are included ($8\cdot 8$) ?
- For other modulation methods such as 16-QAM, where is the 16 coming from?