In modern digital telecommunication methods, a sinewave signal, at a single frequency, is passed through a simple phase shift network to shift it's phase by 90 degrees. The result is two signals. One that is "in phase" (the "I" signal) and the other that is "out of phase" by 90 degrees (the "Q" or quadrature signal). When these two signals are independently modulated in amplitude (think AM), and then both are sent to a suitable mixer, "magic" happens. That "magic" is basically the "drawing" of what is effectively an X-Y graph, based on phase and amplitude (i.e. the amplitudes of the I and Q signals), where the values of each point can be positive or negative. When these signals are demodulated at the other end of the communications link, a constellation is effectively "seen", whose complexity depends on how many amplitude levels are encoded by the modulator. Each "instant" in the modulation timeframe, sees a single point being "drawn" in the X-Y graph, like the animation below. One way to help to imagine this, is to think of there being 16 LEDs in a 4x4 matrix, and only one LED lights up at each instant. Each of those states, or points drawn on the graph, is a so-called symbol (hence the so-called symbol rate in digital signals). In the modulation scheme shown in the animation, it's possible to see that 16 individual states can be described or defined, so this is what is called 16QAM. Those 16 states effectively allow one hexadecimal character (0 to F) to be sent for each symbol instant or state i.e. 4 bits per symbol. This compares to 1 bit per symbol for simple on/off keying of a signal.I'm sorry; I'm radio amateur from the 1960's ( and a chemist by profession). Your assumption that I've ever previously heard of BER was.... in error.
No doubt, you'll enlighten me / us?
The bit error rate (BER) can be estimated if an error correction scheme is implemented by the modulator (it almost always is). It might be a simple parity check, or a checksum value added to the data stream sent to the modulator. The demodulator at the other end of the link, takes the I and Q signals from the receive mixer and, by design, decodes the "value" that it "thinks" it sees for each symbol, and is able to calculate an estimate of the bit error rate. Real life constellations are not full of clear clean points, like the animation. Instead, the "points" are noisy, or fuzzy, to various degrees, and the whole constellation of points can spin, and grow and shrink. Digital modems, and their error correction schemes, are usually implemented using fiendishly clever code running on a dedicated processor.
The diagrams above are from: https://en.wikipedia.org/wiki/Quadrature_amplitude_modulation
Last edited: