Data attribute description has a vital role in the physical significance and intelligent processing of sensing data and improves the usefulness of data effectively. It is an important basis of heterogeneous data sharing and integrating and heterogeneous resource interaction. However, with the development of IoT technology, established descriptions for sensing data attributes are specialized for specific purposes. The physical significance of sensing data is designed for specific applications. The expression of data significance is limited and the structure of data attribute descriptions is strong, which is difficult to achieve the extension of data attributes and reuse of data. In cross-domain applications, this hinders further data processing, which is strongly required in the IoT. To overcome these problems, this paper presents an ontology-based data attribute description method for sensing data processing. In this method, the approach of semantic description for data attributes with a layered ontology architecture, which extracts the data attribute information from the database storing the sensing data, allows the data attributes to be independent of specific applications. In the process of data access, the sensing data can be semantically annotated with data attributes for the requirements of specific applications. Further, extension of the data attributes can be achieved by reasoning technologies. This manner of semantic annotation for data unifies the semantic description of the data attributes and creates flexible data schemata. This method can effectively improve the physical significance and useful value of sensing data and can provide computers a more in-depth understanding and facilitate more intelligent data processing of the sensing data. The sharing and integrating of heterogeneous data and interacting of heterogeneous resources can be implemented based on this method.