1. Introduction

Open Forward Error Correction (OFEC) is a block-convolutional code built from extended Bose-Chaudhuri-Hocquenghem BCH(256,239) component codewords, decoded by an iterative soft-decision decoder, and specified identically in ITU-T G.709.6, the Open ROADM Multi-Source Agreement (MSA) W-Port digital specification, the OpenZR+ MSA and the OIF 800ZR Implementation Agreement. With three soft-decision iterations it delivers a net coding gain of 11.1 dB at a post-FEC bit error ratio (BER) of 10^-15 for dual-polarization quadrature phase shift keying (DP-QPSK) and 11.6 dB for dual-polarization 16-state quadrature amplitude modulation (DP-16QAM), correcting a pre-FEC BER threshold of 2.0 x 10^-2 (standard-specified, Open ROADM MSA W-Port B400G digital specification).

That single paragraph carries the two facts an operator buys. The first is the gain: 11.1 dB of net coding gain is roughly the difference between a 400 Gb/s wavelength that stops at a data-centre campus and one that crosses a regional network. The second is the word identical. Before OFEC, every vendor's soft-decision code was proprietary, which meant a line-side wavelength terminated on the same vendor's card at the far end. OFEC is published in full — parity check matrix, interleaver tables, scrambler seed, frame alignment word, and a normative set of test vectors — so a module from one supplier decodes a wavelength launched by another.

The code sits in a specific place in the transmit chain. A Flexible Optical Transport Network (FlexO) information structure arrives from the client mapping layer, a 32-bit cyclic redundancy check (CRC32) and all-zero padding align it to the coder block group boundary, a frame-synchronous scrambler removes payload patterns, four parallel encoders add parity, two stages of interleaving spread burst errors, and a symbol mapper places the coded bits onto the in-phase and quadrature rails of both polarizations. Each of those stages has exact bit counts in the specification, and this article works through them rather than around them. The coherent DSP architecture walkthrough covers the receive-side blocks that feed soft information into the decoder.

Three audiences get different value from what follows. An engineer new to coherent transport can read sections 2 through 4 as a definition of what an FEC block actually contains. A transport designer will use sections 9 and 10, where the rate arithmetic and the reach and margin trade-offs live. An operations engineer will find sections 11 and 12 most directly useful, because that is where pre-FEC BER thresholds, effective signal-to-noise ratio (eSNR) reporting, error marking and link degrade signalling are set against real specification values. For a shorter orientation to the same subject, the earlier MapYourTech treatment of Open Forward Error Correction fundamentals is a good starting point.

2. Coding Terminology and Net Coding Gain

Net coding gain (NCG) is the reduction in required signal-to-noise ratio at a stated reference output BER, corrected for the noise increase caused by the bandwidth expansion that the FEC parity bits force. Coding gain without that correction overstates an out-of-band code, because the added parity widens the receiver noise bandwidth. ITU-T G.Sup39 defines both terms and gives the closed-form expression used throughout this article.

2.1 The Four Quantities That Define a Code

A coherent FEC is fully described, for planning purposes, by four numbers. Code rate R is the ratio of information bits to total transmitted bits; for OFEC it is 111/128 = 0.867, which is stated as 15.3% overhead when expressed as parity relative to information (17/111). Pre-FEC BER threshold is the input error ratio at which the decoder still reaches the reference output BER; for OFEC that is 2.0 x 10^-2. Reference output BER is the post-decoder target, 10^-15 for the OFEC figures. Net coding gain then follows from the first three.

Two further properties matter in operation but do not appear in the headline figure. Latency is set by the block size and the interleaver depth; the Open ROADM specification bounds combined encoder and decoder latency at less than 2 microseconds at 800G. Burst tolerance is set by the interleaver, and is quoted as a maximum correctable burst length of 2,681 bits with a hard decoder.

2.2 Hard-Decision and Soft-Decision Decoding

Hard-decision decoding quantizes each received bit to one or zero before the decoder sees it. Soft-decision decoding passes a multi-level confidence value — in practice a log-likelihood ratio — so the decoder knows which decisions were marginal. On an additive white Gaussian noise (AWGN) channel the difference is worth more than 1 dB at the same redundancy: at 15% redundancy and a reference BER of 10^-15, ITU-T G.Sup39 gives a theoretical hard-decision bound of 11.0 dB and a soft-decision bound of 12.2 dB. OFEC is a soft-decision code and reaches 11.1 dB in the DP-QPSK case, which places it about 1.1 dB from the soft-decision bound at its own redundancy.

That gap is not waste. It is the price of a decoder that fits inside the power and area budget of a quad small form-factor pluggable double density (QSFP-DD) module. Closing the last decibel would need a longer iteration schedule, more soft bits per sample, and more memory, all of which cost milliwatts on a faceplate that has none to spare on a faceplate whose thermal envelope is already fixed by the host line card.