Also known as: M-ary FSK, MFSK, multi-tone FSK
M-ary FSK (MFSK) generalises frequency-shift keying to M distinct tones, so each transmitted symbol selects one of M frequencies and carries log₂M bits.1 Binary FSK is the M = 2 case; 4FSK is M = 4 (2 bits per symbol); systems with 8, 16, or more tones push further. Unusually among modulations, MFSK gets more power-efficient as M grows — at the cost of proportionally more bandwidth.
How it works
The transmitter assigns each group of log₂M bits to one of M frequencies and emits that tone for a symbol interval. If the tones are spaced so they are mutually orthogonal — separated by an integer multiple of the symbol rate — a receiver can detect them non-coherently with a bank of matched filters or an FFT, picking whichever tone has the most energy. No carrier-phase recovery is needed, which makes MFSK robust on fading and Doppler-affected channels.
The efficiency story is the reverse of most schemes. As M increases, the tones become an ever-better approximation of orthogonal signalling, and the energy per bit needed to hit a given error rate falls, approaching the ultimate Shannon limit for large M. The price is bandwidth: doubling the bits per symbol by doubling M multiplies the number of tones, so the occupied spectrum grows roughly linearly with M while the data rate grows only logarithmically. MFSK therefore suits power-limited, bandwidth-rich channels — the opposite regime from spectrally efficient QAM or high-order PSK.
Variants
Land-mobile radio uses the low-M, continuous-phase end of the family: 2FSK and 4FSK shaped for a constant envelope and narrow channel. HF and weak-signal amateur data modes sit at the high-M, orthogonal end, where dozens of tones buy resilience against ionospheric fading. Coherent MFSK and continuous-phase MFSK (CPFSK) are refinements that trade a little of the non-coherent simplicity for better spectrum or distance.
Relevance to SDR
Because MFSK is easy to detect with an FFT, it is a natural fit for software radio and appears widely: 4FSK voice systems (DMR, NXDN, P25 C4FM), multi-tone HF keyboard modes, paging, and various telemetry links. On a spectrogram, high-M MFSK is unmistakable — a staircase of narrow tones hopping between fixed frequencies.
GopherTrunk works at the low-M end: its C4FM/4FSK demodulator recovers the dibit stream for P25, DMR, and NXDN. The broader M-ary FSK theory is documented here to place that 4-level modulation within the larger orthogonal-signalling family and to explain why FSK-based systems tolerate cheap, efficient transmitters.
Sources
-
Multiple frequency-shift keying — Wikipedia, for the M-tone definition, orthogonal spacing, and the bandwidth-versus-power trade-off. ↩