Big sensor data generated by networked sensors poses challenges for processing due to its volume and velocity. Cloud computing provides scalable computing resources to address this. Existing techniques don't efficiently detect and locate errors in large sensor datasets. The paper proposes an approach exploiting cloud platforms and sensor network topology. It classifies error types and analyzes network features to detect errors in temporal or spatial data blocks rather than whole datasets. Detection tasks are distributed across cloud resources to speed up the error detection process while maintaining accuracy. Experiments on a cloud platform demonstrate the approach significantly reduces error detection time in large sensor datasets.