We currently work on a new feature of Qiqbus: a data quality assessment service for IoT data streams through probabilistic and machine learning approaches.
Our Qiqbus feature shall dictate to the field operator which sensors produce low-quality data, that is which ones might degrade the performance of end-user’s applications or services, aiding the field operator to undertake corrective actions (e.g. replace specific sensors with new ones or update the sensor topology in case that data are lost or delayed due to networking issues) in order to overall improve the performance of their data-driven applications and services.
To aid non expert in AI end users, our feature encompasses intuitive map-based User Interfaces, which help end-users to identify easily possible flaws within their sensor substrate.
The diagram depicts an image taken from of our laboratory prototype for outlier detection over timeseries based sensor data. This work has been performed within the scope of our Open Call project with FIESTA-IoT H2020 .
The figure displays outliers caught by Modio’s custom outlier detection approach based on Long Short-Term Memory implementation of Recurrent Neural Networks (LSTM-RNN).
Apart LSTM-RNN method, the system has been applied to find sensors reporting anomalous data, using additionally an efficient novel algorithm developed by Dr Grigorios Loukides (King’s College London) who collaborated with us in our in-house MVP development.
More methods for outlier detection are available within Qiqbus dev repository. Please drop us an email at info@lamdanetworks.io to obtain more information.