Performance Analysis of Direct-Learning Digital Predistortion With Loop Delay Mismatch in Wideband Transmitters

Orthogonal frequency-division multiplexing (OFDM) modulation, which is commonly used in modern digital communication systems due to its flexibility and high spectral efficiency, is sensitive to transmitter nonlinearities introduced by the power amplifiers (PAs). As a key technology in wireless transmitting systems, digital predistortion (DPD) is widely applied to compensate for the transmitter nonlinearities and ensure high power-added efficiency in base stations and mobile devices for cellular systems. In a DPD architecture, the transmitter input and output signals are sampled, aligned, and processed to extract the DPD parameters that would be used to predistort the source signal before transmission to counteract the transmitter nonlinearities. However, the alignment accuracy in DPD is limited by nonideal electronic components and the associated circuitry, which introduces unknown loop delay mismatch and, thus, degrades the overall linearization performance. In this paper, we analyze the effect of a loop delay mismatch on PA model identification and, hence, on DPD linearization performance of the wideband OFDM signals. The expression and upper limit for the normalized mean square error (NMSE) are derived in terms of the loop delay mismatch, transmission signal bandwidth, and signal-to-noise ratio (SNR) of the feedback channel. The expression for the adjacent channel leakage ratio (ACLR) performance is also derived to predict the DPD linearization performance with the presence of the loop delay mismatch. The theoretical analysis reveals that the performance degradation increases with the bandwidth of the OFDM signal and the modulus of the loop delay mismatch. Experiments are performed on the Long-Term Evolution (LTE) Advanced signals with bandwidth ranging from 20 to 100 MHz to evaluate the degradation effect of the loop delay mismatch on the NMSE and ACLR performances.