Correlative Coding

In digital baseband transmission, achieving high spectral efficiency while managing intersymbol interference (ISI) is a fundamental challenge. Correlative coding, also called partial response signaling, is a technique that deliberately introduces a controlled and known amount of ISI to achieve Nyquist-rate signaling within a reduced bandwidth. Unlike conventional systems that try to eliminate ISI entirely, correlative coding exploits it in a structured way, making it possible to transmit at the Nyquist rate using a bandwidth that is half of what a standard system would require.

Core Concept of Correlative Coding

The Nyquist criterion states that to transmit at a symbol rate of 1/T symbols per second without ISI, the minimum required bandwidth is 1/(2T) Hz. In practice, achieving this minimum bandwidth requires an ideal brick-wall filter, which is physically unrealizable. Correlative coding takes a different approach. Instead of fighting ISI, it introduces ISI in a controlled and deterministic way between adjacent symbols, allowing the system to operate at the Nyquist rate within a realizable bandwidth.

The most widely studied form of correlative coding is duobinary signaling. In a duobinary system, the output sample at time nT is defined as y[n] = x[n] + x[n-1], where x[n] represents the current input symbol and x[n-1] is the previous symbol. If the input is binary with levels +1 and -1, then the duobinary output y[n] can take three values: +2, 0, or -2. This three-level output is the direct consequence of the intentional ISI between two consecutive symbols.

The key advantage of this arrangement is spectral. The transfer function of the duobinary filter H(f) = 1 + e^(-j2πfT) has a raised cosine-like shape that smoothly rolls off to zero at f = 1/(2T). This means the system uses only half the bandwidth of a conventional binary NRZ system transmitting at the same rate, while still maintaining a symbol rate of 1/T. The price paid is the three-level output, which has slightly reduced noise margin compared to a two-level system.

Mathematical Expression

The duobinary encoding rule is expressed as a simple recursive or FIR relationship. For a binary input sequence x[n] taking values from {+1, -1}, the output is given by y[n] = x[n] + x[n-1]. The transfer function of this operation in the z-domain is H(z) = 1 + z^(-1), and in the frequency domain H(f) = (1 + e^(-j2πfT)). Taking the magnitude, |H(f)| = 2|cos(πfT)|, which smoothly decreases from 2 at f=0 to 0 at f=1/(2T). This is why duobinary naturally fits within a bandwidth of 1/(2T) Hz.

At the receiver, a simple decision rule is applied. If y[n] = +2, it is inferred that x[n] = x[n-1] = +1. If y[n] = -2, then x[n] = x[n-1] = -1. If y[n] = 0, one of the two adjacent symbols was +1 and the other was -1. However, this direct decoding suffers from error propagation because a wrong decision on one symbol affects the next. This is solved by using a precoder at the transmitter that removes this dependency and allows symbol-by-symbol detection at the receiver.

The precoding operation converts the input data bits d[n] into a precoded sequence x[n] using modulo-2 logic: p[n] = d[n] XOR p[n-1]. The duobinary encoded output is then formed from p[n]. At the receiver, the decoded bit d[n] is obtained simply by checking whether y[n] = 0 (decoded as 1) or y[n] is nonzero (decoded as 0), depending on the precoder convention used. This eliminates error propagation entirely.

Practical Understanding

Correlative coding finds real application in systems where bandwidth is at a premium. In optical fiber communication, duobinary modulation is used to pack more information into limited channel bandwidth while making use of the inherent low-pass filtering nature of optical links. In magnetic recording and hard disk drives, a variant called modified duobinary (y[n] = x[n] - x[n-2]) is used because it suppresses the DC component of the signal, which is important for systems with AC-coupled channels.

The fundamental engineering tradeoff is clear. Correlative coding reduces the required bandwidth by 50 percent compared to conventional signaling at the same data rate. However, the output signal has more levels (+2, 0, -2 in the duobinary case), which means each decision level has less noise margin. Specifically, the minimum distance between levels is halved compared to a binary system. For GATE purposes, this tradeoff between bandwidth and noise margin is the key point to retain.

Solved Numerical Example

Consider a binary sequence transmitted at a rate of 4000 symbols per second using duobinary correlative coding. The precoded binary symbols are x = [+1, +1, -1, +1, -1, -1]. The duobinary output sequence is formed by y[n] = x[n] + x[n-1]. The required transmission bandwidth for this duobinary system is 1/(2T) = Rb/2, compared to Rb for conventional NRZ binary signaling.

Given:
Symbol rate Rb = 4000 symbols/s
Symbol period T = 1/4000 = 0.25 ms
Input precoded sequence: x = [+1, +1, -1, +1, -1, -1]

Why this formula applies:
Duobinary rule y[n] = x[n] + x[n-1] introduces controlled ISI between adjacent symbols.
Assume x[-1] = 0 (initial condition).

Formula:
y[n] = x[n] + x[n-1]

Substitution:
y[0] = x[0] + x[-1] = +1 + 0 = +1
y[1] = x[1] + x[0]  = +1 + 1 = +2
y[2] = x[2] + x[1]  = -1 + 1 =  0
y[3] = x[3] + x[2]  = +1 - 1 =  0
y[4] = x[4] + x[3]  = -1 + 1 =  0
y[5] = x[5] + x[4]  = -1 - 1 = -2

Calculation:
Required bandwidth (Duobinary) = 1/(2T) = 4000/2 = 2000 Hz
Required bandwidth (Standard NRZ) = 1/T = 4000 Hz
Bandwidth saving = 50%

Final Answer with units:
Duobinary output sequence: [+1, +2, 0, 0, 0, -2]
Bandwidth required = 2000 Hz (vs 4000 Hz for conventional NRZ)

Exam Tip: In GATE problems on correlative coding, remember that duobinary y[n] = x[n] + x[n-1] produces 3 output levels and occupies bandwidth = Rb/2. Modified duobinary y[n] = x[n] - x[n-2] produces 3 levels but has zero DC component, making it suitable for AC-coupled channels. Do not confuse the two.

Mechanism Explained

• Correlative coding intentionally introduces controlled ISI across adjacent symbols to enable signaling at the Nyquist minimum bandwidth using realizable filters.
• Duobinary rule y[n] = x[n] + x[n-1] converts a 2-level binary input into a 3-level output (+2, 0, -2) and fits within bandwidth 1/(2T) Hz for symbol rate 1/T.
• Modified duobinary y[n] = x[n] - x[n-2] spans 3 symbols, nulls the DC component, and is preferred in AC-coupled systems such as magnetic recording.
• Precoding at the transmitter eliminates error propagation at the receiver by making each symbol decision independent, using modulo-2 encoding.
• The bandwidth saved is exactly 50 percent compared to a standard binary NRZ system carrying the same data rate, which is the primary motivation for using correlative coding in bandwidth-limited channels.

Quick Revision

• Correlative coding = controlled ISI introduced deliberately to transmit at Nyquist rate within half the normal bandwidth.
• Duobinary: y[n] = x[n] + x[n-1]. Output levels: +2, 0, -2. Bandwidth = Rb/2. DC present.
• Modified Duobinary: y[n] = x[n] - x[n-2]. Output levels: +2, 0, -2. Bandwidth = Rb/2. Zero DC component.
• Transfer function: H(z) = 1 + z^(-1) for duobinary. |H(f)| = 2|cos(πfT)| smoothly rolls off to zero at f = 1/(2T).
• Precoding removes error propagation. Without precoder, a single detection error spreads to all subsequent symbols.
• GATE trap: Correlative coding does NOT violate the Nyquist criterion. It uses ISI purposefully within the Nyquist bandwidth, not beyond it.
• Practical applications: Optical fiber duobinary modulation, hard disk drive recording, wireline DSL systems.