Power analysis is one of the most effective techniques in side channel analysis.This technique utilizes the power consumption that are relative to the intermediate state of cryptographic algorithm to recover the secret information in the cryptographic devices.Differential power analysis is the first method of power analysis and it is one of the most fundamental analysis techniques.However

in practical scenario

the efficiency of differential power analysis is largely affected by the noise of power traces.Consequently

this leads to the low efficiency of differential power analysis and it is hard to recover the secret key.To address this issue

a new method that is based on singular value decomposition to select power traces is proposed.The power traces of high quality can be selected when this method is applied to improve the efficiency of differential power analysis.The experiments verify the validity of the method.Further

the experimental results show that our method is much better compared with the existing method.Using the method of this paper only 124 power traces is needed to achieve the success rate of 80%

while the normal differential power analysis needs 490 power traces.In addition

when analyzing the data of DPA Contest V1

using the method of this paper only 53 power traces is needed to achieve the success rate of 80%

while the normal differential power analysis needs 195 power traces.Two experiments on different subjects verify the effectiveness of our method.