Glossary term

Modulation Order

Engineering definition of modulation order covering constellation size, bits per symbol, symbol rate, spectral efficiency, SNR, EVM and validation tradeoffs.

Branch: Telecommunications Engineering
Glossary type: quantity
Content: Glossary term
Updated: Jun 26, 2026
Revision: v1.0.0 · reviewed

Definition

quantity

Modulation order is the number of distinct symbols or constellation states available in a digital modulation format.

Modulation order is usually written as M. It controls how many coded bits can be mapped onto one modulation symbol through k = log2(M). Higher modulation order can reduce symbol rate or increase spectral efficiency for the same coded bit rate, but it places constellation points closer together and usually requires higher SNR, lower EVM, cleaner carrier recovery, better timing, lower phase noise and more linear RF hardware.

Modulation order is the number of distinct waveform states in a digital modulation format. It is normally written as M. QPSK has M=4, 16-QAM has M=16, and 64-QAM has M=64.

The order determines how many coded bits one symbol can carry. It also determines how close the constellation points are for a given signal power. This makes modulation order a central design variable in throughput, occupied bandwidth, receiver sensitivity, EVM, linearity and adaptive modulation.

Bits Per Symbol

For a modulation alphabet with M states:

k=\log_2(M)

where k is coded bits per symbol. Conversely:

M=2^k

For common square-QAM and PSK modes:

QPSK: M=4, k=2;
16-QAM: M=16, k=4;
64-QAM: M=64, k=6;
256-QAM: M=256, k=8.

Symbol Rate Link

At the mapper boundary, coded bit rate and symbol rate are related by:

R_{b,coded}=kR_s

so:

\displaystyle R_s=\frac{R_{b,coded}}{\log_2(M)}

Raising modulation order can reduce symbol rate for the same coded bit rate. That can reduce occupied bandwidth for a given pulse shape, but only if the channel and hardware can support the denser constellation.

Spectral Efficiency Link

With code rate R_c, overhead fraction alpha_oh and roll-off factor alpha, a first-pass payload spectral-efficiency screen is:

\displaystyle \eta_p\approx\frac{R_c\log_2(M)}{(1+\alpha_{oh})(1+\alpha)}

This formula explains why modulation order, code rate and overhead must be stated together. A high M with heavy coding, pilots, retransmissions or guard overhead may deliver less useful efficiency than a lower-order mode in a clean accounting model.

Constellation Spacing

For square QAM with the same average symbol energy, increasing M reduces nearest-neighbor spacing. A simplified spacing relation is:

\displaystyle d_{min}\approx\sqrt{\frac{6E_s}{M-1}}

where E_s is average symbol energy. The ratio between 64-QAM and 16-QAM spacing at the same E_s is:

\displaystyle \frac{d_{64}}{d_{16}}=\sqrt{\frac{16-1}{64-1}}=0.49

That smaller decision distance is why higher-order modulation is more sensitive to noise, interference, phase error, IQ imbalance and nonlinear compression.

SNR And EVM Consequences

Higher modulation order usually requires a lower error-vector magnitude and a higher usable SNR or SINR. If the receiver constellation is blurred by thermal noise, phase noise, carrier frequency offset, timing jitter, IQ imbalance, multipath or adjacent-channel interference, symbols cross decision boundaries more often.

A practical mode decision therefore uses margin:

M_{SNR}=SNR_{usable}-SNR_{required}

and may also enforce an EVM limit:

EVM_{meas}\leq EVM_{limit}

The required values are implementation and standard dependent. They should come from the selected MCS table, lab characterization or field validation data, not from modulation order alone.

Worked Example

A link must deliver:

R_p=80\ \text{Mbit/s}

with overhead:

\alpha_{oh}=0.15

and code rate:

R_c=0.75

The coded bit rate is:

\displaystyle R_{b,coded}=\frac{80(1+0.15)}{0.75}=122.7\ \text{Mbit/s}

With roll-off:

\alpha=0.25

the approximate bandwidth for QPSK is:

\displaystyle R_s=\frac{122.7}{2}=61.3\ \text{Msymbol/s}

B\approx61.3(1.25)=76.7\ \text{MHz}

For 16-QAM:

\displaystyle R_s=\frac{122.7}{4}=30.7\ \text{Msymbol/s}

B\approx30.7(1.25)=38.3\ \text{MHz}

For 64-QAM:

\displaystyle R_s=\frac{122.7}{6}=20.4\ \text{Msymbol/s}

B\approx20.4(1.25)=25.6\ \text{MHz}

The corresponding payload spectral-efficiency screens are about:

\displaystyle \eta_{QPSK}=\frac{80}{76.7}=1.04\ \text{bit/s/Hz}

\displaystyle \eta_{16QAM}=\frac{80}{38.3}=2.09\ \text{bit/s/Hz}

\displaystyle \eta_{64QAM}=\frac{80}{25.6}=3.13\ \text{bit/s/Hz}

Assume measured usable SINR is:

SINR_{usable}=18\ \text{dB}

and the release thresholds are 7 dB for QPSK, 16 dB for 16-QAM and 24 dB for 64-QAM. The margins are:

M_{QPSK}=18-7=11\ \text{dB}

M_{16QAM}=18-16=2\ \text{dB}

M_{64QAM}=18-24=-6\ \text{dB}

64-QAM is the most spectrum-efficient option, but it fails the SINR margin. 16-QAM is the highest-order mode that passes this simple release screen.

Adaptive Modulation

Adaptive systems change modulation order as channel quality changes. A good controller does not switch only on instantaneous SNR. It uses hysteresis, packet-error rate, EVM, channel-estimation confidence, mobility, interference history, queue state and service requirements.

Without hysteresis, the link may flap between modes. Without validation, a mode table may look efficient in a spreadsheet while producing unstable throughput in the field.

Common Mistakes

Do not call modulation order a bit rate. It is a constellation-size parameter. Bit rate also depends on symbol rate, coding, overhead and scheduling.

Do not select the highest M just because it improves a spectral-efficiency calculation. Higher order may reduce bandwidth but increase receiver failures, retransmissions, latency, adjacent-channel sensitivity and power-amplifier linearity burden.

Validation Evidence

A defensible modulation-order decision should state:

selected modulation family and M;
bit mapping and coding rate;
symbol rate and bandwidth convention;
required SNR or SINR by mode;
measured EVM, phase noise and frequency offset;
interference and fading margin;
packet-error or BER target;
fallback and hysteresis rules.

Modulation order is therefore both a throughput lever and a validation burden. It is only useful when paired with the channel, receiver and service evidence that prove the denser constellation can be decoded reliably.

REF

Disciplines