Linearity enhancement of linear frequency-modulated DFB semiconductor lasers based on smoothing algorithms

Frequency-modulated continuous-wave (FMCW) light detection and ranging (LiDAR) offers considerable potential for high-precision ranging and velocity measurement, yet its accuracy is critically limited by the linearity of the laser frequency modulation. To overcome the poor nonlinearity correction of conventional predistortion methods at sweep transition points, this paper introduces a smooth predistortion iterative algorithm. The algorithm is realized in a frequency-modulation nonlinear correction system built on a field-programmable gate array (FPGA). In contrast to traditional predistortion, the proposed technique enables more convergence iterations, substantially reducing modulation nonlinearity. Experimental results show that after 2700 iterations, laser frequency nonlinearity for up-ramp and down-ramp sweeps dropped from 3.3% and 1.3% to 0.022%, respectively. The corresponding residual nonlinearity, expressed as 1-r², reached 8.458×10^-7 (up-ramp) and 8.4208×10^-7 (down-ramp). Ranging accuracy reached a maximum error of 9 mm. After correction with the two algorithms, the same target was imaged, and image entropy was introduced as the metric for evaluating imaging quality. The target was a hollow diffuse reflection cardboard. The imaging results should present two planes, with an image entropy of approximately 1 bit theoretically. However, affected by noise and ranging accuracy, the actual value will be bigger than 1 bit. Image entropy was 1.4063 bits for the proposed algorithm versus 1.9585 bits for the conventional one, confirming the superior image quality of the proposed approach.