The Privacy Preserving Data Mining technique that is used widely to conserve security of data is a random perturbation method. The original data is modified and many copies are created according to the trust levels in each field. The addition of noise level also varies with each trust level. The amount of noise added to the lower order trust level is less, whereas it is high in the higher order trust level. There is a chance for the hackers to reconstruct the original data with some non linear techniques. This challenge is addressed by the proposal of a novel non linear technique. Here, the noise level of each copy is checked after perturbation of the actual data. If any similarity is found in the noise level, reconstruction of the original data is possible. Therefore, additional noise is included in the perturbed data. This process is repeated till zero percent of similarity of data is attained.