What’s All This ‘Poisson’ Stuff?
The ability to generate errors is a basic feature in most test sets, it can be used to establish if the receiver can detect and signal if errors occur. Of course, the VIAVI ONT has taken this to new levels with our advanced FEC stress tools used to validated and troubleshoot FEC correction logic and associated circuitry.
Another area we at VIAVI have greatly improved is basic error generation. Classic basic testers can ‘flip’ every nth bit to set up an error rate of 1 in n, this is a simple approach but unfortunately is not representative of the real-world error profile. In most cases errors tend to follow Poisson statistics – just like buses – where there are large gaps and then three of four arrive together. An interesting application of Poisson statistics was done by Ladislaus Bortkiewicz when he studied Prussian cavalry officers being kicked to death by their horses, which like buses, also came in bunches. Now, back to the errors.
Of course, over time, the error rate will tend to a 1 in n (typically expressed as a rate of 1 in 10^n as the error range can cover a large magnitude – a good NRZ 10G link may run at better than 1 in 10^15 while a pre-FEC coherent link could be around 1 in 10^3).
A regular 1 in an error does not represent any real-world scenario and may lead to misleading behaviour so it is critical to test with ‘real’ noise profiles – not only to test error detecting circuitry but also to validate the performance monitoring applications which need to integrate the errors over time to validate against thresholds. The comprehensive Poisson error injection application allows testing with full confidence – an important factor in todays advanced high-speed communication systems. It replaces fussy optical attenuators which need constant tweaking to give a set error rate with a stable, repeatable error source which follows real Poisson error distribution and can be scaled over a large magnitude of error rates.
If you have any questions of testing and validation, please let us know. We may not know much about horse kicks, but we do know an awful lot about test & measurement.