A Data Cleansing Approach In Smart Home Environments Using Artificial Neural Networks

A smart home is generally equipped with a set of sensors/devices able to provide intelligent and personalized services to end users. These sensors/devices can sense multiple information related to the physical environment and the residents. This information is then transmitted to a central station for further processing through wireless communication. However, the wireless medium is considered vulnerable and the sensors can fail in providing correct measurements. Moreover, a smart home system should also be able to implement a cleaning system of its sensed data and discard those instances that are erroneous or incoherent. To achieve the data quality improvements, this paper proposes a new approach that uses an Artificial Neural Network (ANN) to detect faulty measurements. The proposed scheme can prematurely and efficiently detect outlier data before forwarding it to a central station. The performance of the solution is validated through simulations, using realistic datasets, and compared with other well-known models. Our findings demonstrate that the proposed approach outperforms the compared models in terms of accuracy, f-score, recall and precision metrics.