Reflectometry in power lines (spread spectrum TDR)

Hi,
I am working on a project using SSTDR to find the impedance mismatches on the electrical wire. I send a PN code signal and record the reflection (after it reaches to the impedance mismatch). Using the cross correlation of reflected signal and incident, I find the delay that represents location of impedance mismatch. But, if there are multiple impedance mismatches along the wire (@ different locations), the middle ones have a tiny reflection and make it hard to locate them.
Does anyone have any idea on how to amplify those reflections through signal processing?
I really appreciate your time in advance.

This is a typical remote-sensing (e.g., radar, sonar), or channel sounding (e.g., training an equalizer for a communication channel) problem. Since you are using SS, you can get more processing gain by increasing the length of the transmitted PN sequences, or improving their autocorrelation properties, if they're not optimal.
Otherwise, a survey of basic radar/sonar/channel sounding principles may provide some ideas.

You can have a long PN sequence.Also you may repeat the same sequence many times(periodic) and find the average.This will perform better against white noise.The smaller echo also will be visible if you average the cross correlation(periodic) many times.

I was going to write that!
Reasons to go to a longer PN sequence that I can see are either because the longest return time is longer than your PN sequence repeat time, or because the response to other reflections is drowning out the reflection you want.
I'm 99.44% sure that a longer PN sequence is going to tend to have a lower floor in its autocorrelation function, so if the problem is self-interference then you couldn't correct it with more repetitions.
If you just need more process gain, and going to a longer PN sequence is cheap, I see no reason not to do just that -- even if your problem is just white noise.

Thank you all for your response and suggestion. I am going to run with a longer PN sequence.