Heterogeneous sensor systems collect a large amount of sensor data. Because of limited processing capability of sensor nodes, such data are typically stored and processed in a centralized environment in the cloud. However, when data are managed and processed by multiple cloud servers, users do not physically possess these data. The risk of unauthorized access to the cloud data increases dramatically. Therefore, enforcing the integrity of the outsourced data remotely becomes a challenge in the heterogeneous sensor system. Remote data possession auditing is a solution to guarantee the availability and integrity of data outsourced to the cloud. Most existing solutions only target at a single cloud service provider (CSP) environment without considering the multi-cloud scenario. How to audit the data possession among the distributed CSP servers effectively becomes a major concern. In this paper, we propose a parallel cloud data possession checking scheme for the multi-cloud environment. Our approach utilizes the homomorphic verification tag created by the Paillier cryptosystem to support unlimited query challenges and introduces the error-correction encoding method to ensure error localization and data correction. We prototype our solution and analyze the scheme comprehensively. Our evaluation results demonstrate the efficiency and security of our method. It also supports simultaneous verification of multiple CSPs. Figure 1. Overall architecture of heterogeneous sensor system.As another example, heterogeneous sensors are also used in earthquake or volcano monitoring, where different types of sensors, including temperature and acceleration, are used to obtain environmental data that are processed in cloud servers [5]. When such systems are integrated with cloud-storage infrastructure, users lose the physical possession of the out-sourced data. There is a dramatic increase in the potential risk in unauthorized access to the data, such as the incident in the S3 service of Amazon [6]. We will present the generalized architecture in Section 3.Intuitively, to ensure that data are stored correctly, we can verify data using general cryptographicbased approaches, for example, hash functions and digital signature schemes. Deswarte et al. [7,8] first make use of the RSA signature scheme combined with hash functions to verify the correctness of outsourced files. However, such a verification scheme needs to access or retrieve the whole outsourced data, which will cause high performance overhead (e.g., computation complexity and communication cost).The work of [9, 10], and [11] proposed the provable data possession model. To reduce the computation and communication cost, these schemes utilize the sampling technique combined with homomorphic verification instead of the whole-file retrieval verification. The approach proposed by Ateniese et al. [9] supports public data possession audit with unbounded times of challenge. Most previous schemes only verify the data correctness without error identification, localization, and correction. ...