Validation is an alternative approach to the establishment of the trust between two or more entities. The validation allows a challenger to evaluate the trustworthiness of an attesting system based on provided evidence. Early attempts show the evidence is always organized in a linear structure or a balanced tree. These methods, however, are not optimal when facing a complex system and privacy issues. In particular, the Stored Measurement Log (SML) to record the execution history of the attesting system brings efficiency, scalability and privacy problems. We attempt to mitigate them through a new algorithm that uses an unbalanced tree to manage the SML. Leaves represent measurement values of components, and the root is protected by platform configuration registers, which is the same as the balanced tree. The location of each node, however, is constantly mutative until the probability distribution of all components tends towards stabilization. The greater the probability of the leaf is, the closer it is to the root, which benefits the validation. We build a prototype program that is developed through the Trusted Platform Module emulator 0.7. We touch on some mechanisms to reduce the size of the tree and avoid privacy leak. Finally, we perform an in-depth analysis of the validation efficiency, and present the impression of the time complexity of the measurement. Results show that the validation obtains the logarithmic speed-up (O((n/N)lb(n))) and that the time complexity of the measurement process is also acceptable. Besides, our scheme demonstrates advantages in privacy protection and scalability.