To address the problems of high noise and low signal-to-noise ratio in singleended Brillouin optical time-domain analysis systems, a denoising method based on empirical mode decomposition (EMD) was developed. Theoretical analyses of the denoising principle of EMD and the sensing principle of single-ended Brillouin optical time-domain analysis in a fewmode optical fiber were conducted. The denoising effect of EMD was comparatively analyzed by constructing a single-ended Brillouin optical time-domain analysis temperature-sensing system.The experimental and simulation results show that the EMD algorithm exhibited a good denoising effect on the temperature sensing system and that it improved the signal-to-noise ratio and temperature measurement accuracy by approximately 3.06dB and 0.98℃, respectively.