Field Guide · term

Also known as: BER

Bit error rate (BER) is the number of received bits that differ from the bits that were sent, divided by the total number of bits — the single most direct measure of a digital link’s health.1 A BER of 10⁻³ means one bit in a thousand is wrong; a clean link might run 10⁻⁶ or better. BER is what forward error correction exists to reduce, and what every other link metric ultimately tries to predict.

10⁻¹ BER 10⁻⁶ Eb/N0 (dB) → uncoded coded coding gain
A BER waterfall curve: error rate plunges as Eb/N0 rises. Coding shifts the curve left — the horizontal gap at a target BER is coding gain.

How it works

BER is estimated by comparing received bits against a known transmitted sequence and counting mismatches over a long enough run to be statistically meaningful — to confirm a BER of 10⁻⁶ with confidence you must observe on the order of 10⁷ or more bits. For a given modulation, BER is a smooth function of the per-bit signal-to-noise ratio, expressed as Eb/N0. Plotting BER (log scale) against Eb/N0 (dB) yields the characteristic waterfall curve: nearly flat and high at low SNR, then plunging almost vertically once enough energy per bit is present. The steepness is why digital links have a “cliff” — a couple of dB can move a channel from unusable to flawless.

Different modulations trace different waterfalls. Robust binary schemes like BPSK tolerate low Eb/N0; dense QAM constellations pack more bits per symbol but need higher Eb/N0 for the same BER because their points sit closer together.

Variants

  • Uncoded (raw) BER — measured on the demodulated bitstream before decoding. This reflects the channel and modulation directly.
  • Coded (post-FEC) BER — after error correction. FEC shifts the waterfall left; the horizontal distance to reach a target BER is the coding gain, typically several dB.
  • Symbol / frame / block error rate — related metrics counting whole wrong symbols, frames, or blocks rather than individual bits, often more meaningful for packet systems.
  • BER floor — an irreducible error rate that persists no matter how high the SNR climbs, caused by implementation flaws such as ISI, phase noise, or timing jitter.

In practice

  • Standards specify a target BER (or residual BER after FEC) as the pass/fail criterion for receiver sensitivity — for example a BER of 5% is a common P25 sensitivity reference point.
  • Soft-decision decoding lowers BER for the same channel by feeding the decoder confidence values instead of hard 0/1 calls, typically buying ~2 dB over hard decisions.

Relevance to SDR

BER is the honest bottom line for any digital decoder. In trunked systems — P25, DMR, NXDN, TETRA — a rising BER on the control channel means missed channel grants and lost calls long before voice becomes unintelligible. GopherTrunk does not usually surface a raw BER number to users, but the same physics governs its decode chain: as demod SNR falls, bit errors climb, the FEC eventually cannot keep up, and frames fail their CRC checks. Understanding the waterfall explains why marginal signals fail abruptly rather than gracefully — GT is riding the cliff.

Sources

  1. Bit error rate — Wikipedia, definition, measurement, and the BER-versus-Eb/N0 relationship. ↩

See also