Apart from FSK, PSK (phase shift keying) has become most popular & now the mainstay of systems. In this, the phase of a Sine wave (of constant amplitude & frequency as there should be no ambiguity that would interfere with phase information, ϕ) carrier is varied, one phase to represent a "0" & the shifted phase as "1". This two level shift is called Bi-PSK or BPSK. Sine wave with ϕ=0º (unchanged) is the 0 bit while ϕ=180º or π Radians, is 1 bit. At the receiving end, there is a phase discriminator (now a simpler design in which the incoming signal is matched against a combination of Cosine & Sine waveforms generated in a very high Frequency Synthesizer). If the phase is same it outputs a 0 bit and if it isn't it outputs 1 bit.
Two bits are combined too and that results in a combination of 4 or 2²: [00], [01], [11], [10], each combination assigned to a particular phase like
[00] = 45º or π/4,
[10] = 135º or π -π/4,
[11] = 225º or π+π/4,
[01] = 315º (= -45º) or 2π -π/4 (= - π/4).
The circuitry renders the phase transition at the transmitting & the receiving end.
It is called QPSK (Quadrature PSK) when the phases are equi-spaced at π/2 Rad (= -90º).
8 PSK is achieved when 2³ levels of discrete phase levels are assigned spaced π/4 Rad apart.
16 PSK (2⁴ to bunch 4 bits as a "symbol"), ... 256 PSK (2⁸) are used even.
You must have noticed that the bits are bunched into k bits where the number of discrete levels of phase is 2ᵏ. At each increase in k the symbol packs double in number for the same carrier frequency that translates to halving the Band Width requirements which is the secret of a m-PSK system.
A 256PSK can reduce the BW (per symbol) requirement to 1/8 ͭ ͪ of the BW, as 8 bits are bunched into a symbol for 1 phase shift.
Yet if you view the diagram (of phases in a Polar representation), called "constellation" chart you'll notice that the points on the periphery of a unit circle fall too close & errors creeping in propagation, might bring two adjacent symbols close enough to have "inter-symbol interference " with one symbol being interpreted as the next one by the receiver. Only precision can save it from error; but there is a limit till which the system designer can go.
So, instead of placing the points or symbols (on constellation diagram) on a unit circle, they are stacked as points in a square array or pattern. It is another scheme, involving the symbols assigned at specific Amplitudes in addition, as
A Cosϕ+B Sinϕ.
{ There are many manufacturer/vendors who can sell solutions to one's requirement of digital communications, for Satellite Communications. I noticed that "Qualcomm" is the leader }