Definition
Bit error rate, known by its abbreviation BER, is the number of errored bits divided by the total number of bits observed. It is a dimensionless ratio, usually expressed in scientific notation because the values of interest are very small. It is interpreted as the probability that any given bit arrives wrong.
What it depends on
- Signal-to-noise ratio: the higher the ratio, the lower the probability that noise pushes a symbol toward the wrong decision.
- Modulation scheme: schemes with more states per symbol have states that are closer together and need a higher signal-to-noise ratio for the same BER.
- Distortion and interference: they reduce the decision margin even with a good signal-to-noise ratio.
- Synchronization: phase or clock errors in the receiver increase the error rate.
The curve of BER versus signal-to-noise ratio has a waterfall shape: it drops quickly as the ratio improves. A small change in signal can move the rate by several orders of magnitude.
How it is measured
A generator sends a known bit sequence, normally pseudorandom, and a test receiver compares it bit by bit with the expected one. To measure very low rates, many bits must be observed, because with few errors counted the estimate is unreliable. This means that measuring low BER values takes time.
Before and after error correction
Modern systems add error correction codes. It is therefore worth distinguishing between the rate before correction, which reflects the quality of the channel, and the rate after correction, which reflects what the application sees. A high rate before correction may be acceptable if the code reduces it enough, although the available margin is smaller.
How to interpret the result
A BER does not by itself say why there are errors. It is complemented by the constellation, the signal-to-noise ratio and the spectrum. Errors in bursts point to interference or fading; scattered, steady errors point to thermal noise or a modulation that is too dense for the link.