Glossary term

Spectral Efficiency

Engineering definition of spectral efficiency covering payload bit rate, occupied bandwidth, modulation order, code rate, Shannon limits and validation tradeoffs.

Definition

metric

Spectral efficiency is the useful bit rate carried per unit bandwidth at a clearly stated system boundary.

Spectral efficiency measures how much information a communication system moves through a given amount of spectrum or channel bandwidth. It is normally expressed in bit/s/Hz, but the value is only meaningful when the numerator and denominator are defined: gross coded bit rate, net physical-layer bit rate, delivered payload rate, occupied bandwidth, allocated channel bandwidth or regulatory channel bandwidth. High spectral efficiency can improve spectrum use, but it usually requires higher SNR, stronger coding, cleaner synchronization, better linearity and tighter interference control.

Spectral efficiency is the bit rate achieved per unit bandwidth. It is usually reported in bit/s/Hz, but the number is not self-describing. A value based on coded physical-layer bits can be much larger than the value based on delivered payload, and a value based on occupied bandwidth can differ from one based on allocated channel bandwidth.

The metric is useful because spectrum and channel bandwidth are constrained engineering resources. It is also easy to misuse. A high spectral-efficiency target may force a communication link toward higher modulation order, less redundancy, more linear transmitters, tighter carrier recovery, lower phase noise, better channel estimation or a smaller fade margin.

Basic Definition

At a stated boundary:

\displaystyle \eta=\frac{R}{B}

where:

  • eta is spectral efficiency;
  • R is the bit rate at the chosen boundary;
  • B is the chosen bandwidth denominator.

The boundary must be named. R may be coded bit rate, net information bit rate, payload bit rate or measured user throughput. B may be occupied bandwidth, noise-equivalent bandwidth, allocated channel bandwidth or regulatory channel width.

Gross, Net And Payload Values

Gross coded spectral efficiency uses the coded mapper bit rate:

\displaystyle \eta_{gross}=\frac{R_{coded}}{B_{occ}}

Payload spectral efficiency uses delivered payload rate:

\displaystyle \eta_p=\frac{R_p}{B_{occ}}

Channel spectral efficiency uses the full assigned channel:

\displaystyle \eta_{ch}=\frac{R_p}{B_{ch}}

These three values answer different questions. Gross efficiency helps check modulation and symbol-rate assumptions. Payload efficiency shows how much useful traffic the waveform carries after coding and overhead. Channel efficiency is often the business, licensing or system-capacity number.

Modulation And Code Rate Screen

For M-ary modulation:

k=\log_2(M)

where k is coded bits per symbol. With code rate R_c, overhead fraction alpha_oh and raised-cosine roll-off 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 is a screening equation, not a standard definition. OFDM pilots, cyclic prefixes, guard subcarriers, retransmissions, scheduling gaps, packet headers and adaptive fallback may need separate accounting.

Shannon Limit Context

For an ideal additive-white-Gaussian-noise channel:

C=B\log_2(1+SNR)

so ideal spectral efficiency is bounded by:

\eta\leq\log_2(1+SNR)

The ideal minimum SNR for a target efficiency is:

SNR_{min}=2^\eta-1

Practical links require more SNR than this bound because of coding loss, implementation loss, fading, interference, nonlinear distortion, synchronization error, channel-estimation uncertainty and required availability margin.

Worked Example

A radio service must deliver:

R_p=80\ \text{Mbit/s}

It uses 16-QAM:

M=16

with:

R_c=0.75

Non-FEC overhead is:

\alpha_{oh}=0.15

and pulse-shaping roll-off is:

\alpha=0.25

The coded bit rate is:

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

16-QAM carries:

k=\log_2(16)=4\ \text{bit/symbol}

so the symbol rate is:

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

The occupied-bandwidth screen is:

B_{occ}\approx30.7(1+0.25)=38.3\ \text{MHz}

Payload spectral efficiency over occupied bandwidth is:

\displaystyle \eta_p=\frac{80}{38.3}=2.09\ \text{bit/s/Hz}

The same result follows from the screening relation:

\displaystyle \eta_p\approx\frac{0.75\log_2(16)}{(1+0.15)(1+0.25)}=2.09\ \text{bit/s/Hz}

If the assigned channel is 40 MHz, the channel efficiency is:

\displaystyle \eta_{ch}=\frac{80}{40}=2.00\ \text{bit/s/Hz}

The ideal Shannon SNR for 2.09 bit/s/Hz is:

SNR_{min}=2^{2.09}-1=3.25

or:

SNR_{min,dB}=10\log_{10}(3.25)=5.1\ \text{dB}

That value is only an ideal lower bound. A real 16-QAM, rate-3/4 link normally needs a much larger usable SNR after fade margin, interference margin, EVM limits and receiver implementation loss.

Spectral efficiency is not automatically better when it is higher. A higher value can reduce required bandwidth, but it can also increase outage probability or force expensive RF performance. For a fixed payload rate, engineers may choose lower spectral efficiency to gain link margin, range, availability or tolerance to adjacent-channel interference.

For adaptive modulation and coding, spectral efficiency is a mode-selection output. The mode table should pair each efficiency with required SINR, EVM limit, packet-error target, hysteresis, fallback behavior and measurement uncertainty. Reporting the efficiency without the required channel condition is incomplete.

Common Failure Modes

The most common mistake is mixing boundaries: using payload bit rate in the numerator but coded bandwidth assumptions in the denominator, or comparing occupied-bandwidth efficiency with allocated-channel efficiency. Another error is treating Shannon capacity as a required SNR table. Shannon gives an ideal limit; practical systems require an implementation gap.

Spectral efficiency can also hide overhead. Pilots, training fields, cyclic prefix, guard bands, retransmissions and protocol headers may be essential to robustness even though they reduce payload efficiency. Removing them to improve a spreadsheet value may make the waveform impossible to synchronize or validate.

Validation Evidence

A defensible spectral-efficiency claim should state:

  • numerator boundary: coded, net information, payload or measured throughput;
  • bandwidth denominator: occupied, channel, regulatory or measurement bandwidth;
  • modulation order and code rate;
  • overhead items included and excluded;
  • required SNR, SINR or EVM;
  • test condition, traffic model and uncertainty.

Without those details, bit/s/Hz is only a headline number. With them, spectral efficiency becomes a useful engineering metric for comparing waveform choices, capacity plans, spectrum licenses and receiver-margin tradeoffs.

REF

See also