duobinary

Duobinary is a line code that encodes binary data by representing consecutive ones as alternately positive and negative voltage levels, while zeros remain at zero voltage. This creates a three-level signal (positive, zero, negative) from a two-state input, which gives the scheme its name: "duo" for two, "binary" for the input data. The key advantage is spectral efficiency: duobinary compresses the bandwidth requirement compared to standard non-return-to-zero (NRZ) coding, making it valuable in bandwidth-constrained channels.

In a duobinary stream, the first one is sent as a positive pulse, the next one as negative, the next as positive again, and so on. Zeros pass through as zero volts. This alternation means the signal contains very little low-frequency energy; the spectrum is concentrated around the Nyquist frequency, allowing the signal to fit through narrower channels. For a given data rate, duobinary requires roughly half the bandwidth of standard binary NRZ signaling.

Implementation and constraints

Duobinary encoding is typically implemented using a precoder followed by a digital filter. The precoder converts the binary input into a form suitable for the filter, which then produces the duobinary output. Decoding at the receiver requires threshold detection and error correction logic, since small noise or ISI (intersymbol interference) can flip the sign of a pulse and corrupt the decoded bit. This sensitivity to noise is the main drawback: while bandwidth efficiency improves, noise margin decreases.

Early telephone modems and some data transmission systems used duobinary to squeeze higher data rates through voice-grade circuits limited to roughly 4 kHz. Modern digital subscriber line (DSL) systems, optical transmission, and other applications continue to employ variants of duobinary or related schemes when bandwidth is scarce and signal-to-noise ratio is adequate. It remains a textbook example of how intelligent signal shaping trades noise robustness for spectral economy.