Ultra-wideband systems can be used in indoor wireless personal area network (WPAN) or short-range wireless local area network (WLAN) transmission. Yet owing to the effects of indoor dense multipath, it will cause more power consumption. We usually use Rake receiver to improve system performance. However, we should do some compromise between system performance and the design complexity. Thus, the concept of Selective Rake can be used to substitute for the conventional Rake receiver. Selective Rake receiver uses fewer but more powerful paths instead of using all the paths to raise system performance. Hence, we have to precisely detect the multipath components for best performance. Earlier we use CLEAN algorithm to estimate the multipath components. The CLEAN algorithm can be used in selecting the paths with relatively high energy. But as the impact of frequency selective fading makes the transmitted signal distorted, the CLEAN algorithm no longer applies to this situation. Thus, we use Multiscale CLEAN algorithm instead. Multiscale CLEAN algorithm calculate the value of cross-correlation between the received signal and a set of waveforms, and then choose the higher one as the waveform transmitted. Besides, we use Maximal Ratio Combining to weigh the different paths to get the signal with more power. We represent the signal affected by frequency selective fading by using the second derivatives of Gaussian waveform function with different effective widths of pulse. The waveforms corresponding different effective widths have different spectra which represent the different effects of fading. It is seen that that the multiscale CLEAN has better performance than the CLEAN algorithm with more precise estimation of multipath components. In simulation result, we can figure out path searching using Multiscale CLEAN algorithm is more accurate than using CLEAN algorithm. Even the path with smaller energy gain, using multiscale CLEAN algorithm can search successfully, while CLEAN algorithm cannot do.