Making heterogeneous data sources homogeneous manually and off-line can become a high time-consuming task. This paper presents a software architecture that extends the standardized-based architectures for heterogeneous sensors with components to also support devices and data that are not compliant with standards. The defined architecture is based on Internet of Things (IoT) layered architectures that establish perception, network, middleware, application, and business as main layers. To define the architecture, an architectural framework was used; this framework supports the identification of non-compliant data, providing then a different processing path. This proposed architecture covers a wide spectrum of data interoperability addressing the IoT challenge of «Interoperability and Standardization». The implemented solution proved that the processing time between data acquisition and the feeding of analysis algorithms can be reduced from 100% to approximately to 1% with systems based on the proposed architecture compared with those that manage data manually and off-line.