Authors:
Adriana Lipovac
1
;
Vlatko Lipovac
1
;
Mirza Hamza
2
and
Vedran Batos
1
Affiliations:
1
Dept. of Electrical Engineering and Computing, Univ. of Dubrovnik, Dubrovnik, Croatia
;
2
Dept. of Telecommunications, Faculty of Electrical Engineering, Univ. of Sarajevo, Sarajevo, Bosnia and Herzegovina
Keyword(s):
OTDR, BER, Dynamic Range, OFDM CFO.
Abstract:
Optical time-domain reflectometer (OTDR) has long been and is still considered the main test tool for characterizing fiber optic links, i.e. identify and localize refractive and reflective events such as breaks, splices and connectors, and measure their insertion/return loss. Specifically, sufficient dynamic range and thus alike signal-to-noise-ratio (SNR) enable clear far-end visibility even of long fiber links. Moreover, under such conditions, the highest achievable optical bit-error-rate (BER) floor is to the large extent determined by major reflective events such as the specific trace distortion caused by connectors and splices, each with significant return loss. Realizing this has provided the opportunity window to extend the standard OTDR capabilities list by the appropriate trace postprocessing to predict the BER floor. Accordingly, considering the SNR high, and thereby the inter-symbol interference dominant error generating mechanism, we applied the time-dispersion channel mo
del that determines the BER floor by the rms delay spread of the (fiber) channel power-delay profile. We verified the BER floor prediction in the exemplar practical test situation, by measuring the actual BER on the same fiber link, and found the obtained values well matching the OTDR - based predicted ones. Furthermore, when no dominant reflective events are identified on the OTDR trace, it implies very small time dispersion allowing the OFDM symbol cyclic prefix to always prevent inter-symbol interference. This retains the CFO to solely determine the residual BER floor and vice versa, enabling indirect estimation of CFO-induced phase distortion by simple BER testing. With this regard, we abstracted CFO with the AWGN being justified by the Central Limiting Theorem to enable efficient and quite accurate short-term BER (and so CFO phase error) predictions.
(More)