Inter Symbol Interference

In digital communication systems, transmitted pulses must be detected correctly at the receiver. When the channel has limited bandwidth or multipath effects, pulses spread in time and overlap with adjacent pulses. This overlap is called Inter-Symbol Interference (ISI), and it is one of the primary causes of bit errors in baseband digital transmission.

What Is Inter-Symbol Interference

Every digital symbol occupies a time slot of duration T seconds, where T = 1/Rs and Rs is the symbol rate. For perfect detection, the pulse representing one symbol should have zero amplitude during all other symbol periods. In practice, a bandlimited channel cannot pass infinitely sharp pulses, causing them to spread beyond their allocated slot.

The received signal at any sampling instant contains contributions not just from the intended symbol but also from neighbouring symbols that have leaked in. This leakage is ISI. It adds or subtracts from the true signal value, effectively reducing the distance between signal levels and increasing the probability of error.

ISI is fundamentally a time-domain spreading problem caused by the frequency-domain limitation of the channel. A channel with bandwidth B can support pulses no narrower than 1/(2B) seconds by the Fourier duality principle. If the symbol period T is smaller than 1/(2B), significant ISI results.

Mathematical Expression of ISI

The received signal r(t) in a baseband system can be written as a convolution of the transmitted sequence with the combined impulse response of transmit filter, channel, and receive filter. Let p(t) be the overall pulse shape at the receiver input. The sampled output at time nT is:

y(nT) = a_n * p(0) + sum over k not equal to n of [a_k * p((n-k)T)] + noise

The first term is the desired symbol. The summation is the ISI term. For ISI-free transmission, p(kT) must equal zero for all nonzero integers k. This condition, known as the Nyquist ISI criterion, defines the ideal pulse shape requirement.

Physical Causes of ISI

The primary physical cause of ISI in wireline baseband systems is the bandlimited nature of the transmission medium. Cables, PCB traces, and optical fibres all exhibit frequency-dependent loss, causing high-frequency components of a pulse to be attenuated more than low-frequency components. This produces pulse broadening in the time domain.

In wireless systems, multipath propagation is an additional cause. Copies of the transmitted signal arrive at the receiver with different delays due to reflections from buildings, terrain, and other objects. Each delayed copy acts as a smeared version of the original pulse, causing adjacent symbol periods to interfere with each other.

The severity of ISI depends on the ratio of the channel delay spread to the symbol period. When the delay spread is much smaller than T, ISI is negligible. When it becomes comparable to T or larger, ISI becomes severe and equalisation is necessary.

Practical Understanding and Mitigation

ISI directly degrades the bit error rate (BER) of the system. Even a small amount of ISI can create an error floor, meaning BER does not improve beyond a certain point regardless of how much transmit power is increased. This is a key practical reason why pulse shaping and equalization are non-optional in real systems.

Mitigation strategies include careful pulse shaping at the transmitter using raised cosine filters to satisfy the Nyquist criterion, channel equalization at the receiver to invert the channel distortion, and in wireless systems, orthogonal frequency division multiplexing (OFDM) which converts a wideband channel into many narrowband subchannels where ISI per subcarrier is minimal.

Given:
Symbol rate Rs = 1 Mbps (binary), so T = 1 microsecond
Channel bandwidth B = 400 kHz

Why this formula applies:
Nyquist minimum bandwidth for ISI-free transmission is B_min = Rs/2 = 500 kHz.
Since actual B = 400 kHz < 500 kHz, the channel cannot support the symbol rate without ISI.

Formula:
B_min = Rs / 2

Substitution:
B_min = 1,000,000 / 2 = 500,000 Hz = 500 kHz

Calculation:
Available B = 400 kHz < Required B_min = 500 kHz
Bandwidth deficit = 500 - 400 = 100 kHz

Final Answer:
The channel introduces ISI because it is 100 kHz short of the minimum Nyquist bandwidth.
To avoid ISI, either reduce Rs to 800 kbps or widen the channel bandwidth to at least 500 kHz.

Exam Tip: GATE frequently tests whether a given channel bandwidth is sufficient for a given symbol rate. Remember B_min = Rs/2 for binary NRZ. If channel B is less than Rs/2, ISI is unavoidable without equalization.

ISI Effects Summary

• ISI adds an unwanted voltage offset to the sampled signal, reducing noise margin and increasing BER.
• ISI cannot be removed by increasing transmit power alone since it scales with signal amplitude.
• The minimum bandwidth needed for ISI-free transmission at symbol rate Rs is B_min = Rs/2 (Nyquist limit for ideal sinc pulses).
• In practice, excess bandwidth above Nyquist minimum is used through raised cosine pulse shaping to relax filter requirements while controlling ISI.
• Equalisers at the receiver undo channel distortion to reduce residual ISI in deployed systems.

Quick Revision

• ISI is the leakage of one symbol into adjacent symbol time slots due to pulse spreading.
• Root cause: channel bandwidth is insufficient to pass pulses without time-domain spreading.
• Nyquist formula: B_min = Rs/2 for ISI-free baseband transmission with ideal sinc pulses.
• ISI creates an error floor; increasing power does not eliminate it once it exists.
• Mitigation: raised cosine pulse shaping (transmitter side) and channel equalisation (receiver side).
• GATE trap: confusing ISI with AWGN noise. ISI is deterministic and signal-correlated; AWGN is random and independent.
• Multipath delay spread divided by symbol period T gives a rough measure of ISI severity in wireless systems.