Zero-Forcing Equalizer

A linear equalizer that minimizes the mean squared error between the transmitted and estimated symbols, balancing ISI cancellation with noise amplification. Unlike the Zero-Forcing equalizer, which strictly inverts the channel response, the MMSE equalizer incorporates the signal-to-noise ratio (SNR) into its filter design. This prevents the catastrophic noise enhancement that occurs in deep spectral nulls, making it the preferred linear solution in practical low-SNR environments.

A non-linear equalizer that uses previously detected symbols to estimate and subtract post-cursor intersymbol interference from the current symbol. The DFE consists of a feedforward filter and a feedback filter. By canceling trailing interference without amplifying noise, it significantly outperforms linear equalizers like Zero-Forcing in channels with severe multipath. However, it is susceptible to error propagation if incorrect decisions are fed back into the loop.

An optimal detection strategy, often implemented via the Viterbi algorithm, that considers the entire sequence of received symbols to determine the most likely transmitted bit stream. Unlike the Zero-Forcing equalizer, which makes symbol-by-symbol decisions, MLSE searches a trellis of channel states to find the sequence that minimizes the overall error probability. It provides the theoretical benchmark for equalizer performance but suffers from exponential computational complexity as the channel memory length increases.

A computationally efficient equalization method performed on a block of received symbols using the Fast Fourier Transform (FFT). FDE converts the time-domain convolution of the channel into a simple element-wise multiplication in the frequency domain, drastically reducing complexity for channels with long delay spreads. This approach is the foundation of single-carrier frequency domain equalization (SC-FDE) and is closely related to the equalization used in OFDM systems.

A stochastic gradient descent adaptive algorithm used to iteratively update the coefficients of a linear equalizer. The LMS algorithm adjusts filter taps to minimize the instantaneous squared error between the desired output and the actual filter output. While computationally simple and widely used for adaptive equalization, its convergence speed is dependent on the eigenvalue spread of the input signal's autocorrelation matrix, making it slower than RLS in highly dispersive channels.

A blind adaptive equalization algorithm that exploits the constant envelope property of modulation formats like PSK and QAM to update filter taps without a training sequence. The CMA penalizes deviations from a constant modulus, allowing the equalizer to converge and open the eye diagram without prior knowledge of the transmitted symbols. This is particularly useful in broadcast and interception scenarios where pilot symbols are unavailable, though it suffers from slower convergence than data-aided methods.