With the swift advancement of machine learning technology, the volume of information is expanding at an exponential rate. An increasing number of users choose to store data in cloud servers, so it is crucial to ensure the integrity of data. The provable data possession (PDP) scheme allows it to obtain integrity proofs like hash-based schemes while consuming less bandwidth. However, most existing PDP schemes rely on the third party auditor (TPA) to verify the integrity of the data for users, which may lead to the problem that TPA steals user data or colludes with cloud servers. We proposed a scheme to replace TPA with a trusted execution environment (TEE), forming a system framework consisting of users, cloud servers, and TEE. This system can effectively reduce the user’s communication and computing overhead and improve the credibility of the system model. Our scheme places the two core stages of challenge and verification in a fully trusted environment, which can eliminate the hidden dangers brought by TPA. We use Intel SGX technology to deploy on a real cloud storage platform and apply the Enclave area created by Intel SGX to the PDP scheme. Our experiments show that the performance calculation overhead is reasonable based on the elimination of TPA dependencies.