Timing Recovery

A non-data-aided (NDA) feedback algorithm requiring only two samples per symbol. The error is computed as e[n] = (y[n] - y[n-2]) * y[n-1], where y[n] is the current sample and y[n-2] is the sample from the previous symbol. This algorithm is remarkably robust to carrier phase offsets, making it independent of carrier recovery. It is the standard choice for BPSK/QPSK systems.

A decision-directed (DD) algorithm requiring only one sample per symbol. The error signal is e[n] = (a[n] * y[n-1]) - (a[n-1] * y[n]), where a[n] are the decided symbols. Its primary advantage is computational efficiency at high data rates. However, it is sensitive to carrier phase errors and requires a pre-converged equalizer, making it ideal for high-speed wireline applications.

A classic feedback synchronizer that performs energy detection on early and late samples relative to the current symbol boundary. The error signal is the difference in energy: e[n] = |y_early|^2 - |y_late|^2. When perfectly synchronized, the early and late energies are equal. This method is effective for pulse-shaped signals but requires at least three samples per symbol.

A feedforward architecture that uses a bank of parallel filters to compute intermediate samples between the original ADC samples. The timing error is estimated using a maximum likelihood (ML) criterion, and the correct fractional delay is selected via a Farrow structure or cubic interpolator. This method is highly effective for burst-mode and multi-rate systems.

A feedforward square-law timing estimator designed for burst-mode transmission. It computes the timing offset by evaluating the spectral component at the symbol rate of the squared magnitude of the received signal. The estimate is derived from the phase of the discrete Fourier transform (DFT) at f = 1/T. It requires four samples per symbol and is ideal for rapid acquisition in packet-based systems.

The difference in frequency between the transmitter and receiver local oscillators. This impairment causes a rotating phase error in the received constellation, making symbol detection impossible if not corrected before timing recovery. CFO estimation is often performed jointly with timing synchronization in data-aided algorithms.

An optimal linear filter that maximizes the signal-to-noise ratio (SNR) at the sampling instant. The filter's impulse response is a time-reversed, complex-conjugated copy of the transmitted pulse shape. The output of the matched filter is the signal used by the timing recovery algorithm to find the peak eye opening.

A two-dimensional scatter plot mapping the in-phase (I) component on the x-axis against the quadrature (Q) component on the y-axis. Timing jitter manifests as a cloud of points around the ideal symbol locations. A properly synchronized receiver produces tight, distinct clusters.

A comprehensive metric quantifying the deviation of measured constellation points from their ideal reference positions. Timing error is a direct contributor to EVM degradation. The metric captures the combined impact of phase noise, IQ imbalance, and timing jitter.

An efficient implementation of interpolation and decimation that rearranges filter coefficients to operate at the lower output sample rate. In timing recovery, polyphase filter banks are used to implement arbitrary fractional-delay interpolation for symbol resampling without high computational cost.

The process of digitally translating a sampled bandpass signal to complex baseband using a numerically controlled oscillator (NCO) and decimating filters. Timing recovery operates on the baseband IQ samples produced by the DDC, typically at 2 to 4 samples per symbol.