Design of an Efficient Multidomain Augmented Data Aggregation Model to Solve Heterogeneity Issues for IoT Deployments
Meesala Sravani, Suniti Purbey, Burada Chakradhar, Ashutosh Kumar Choudhary · 2023
Heterogeneity of data is a common issue in data retrieval in Internet of Things (IoT) Networks, when dealing with data from multiple sensors & sources. To retrieve data from heterogeneous sources, there are several techniques proposed by researchers, which include, Data integration, Data federation, Semantic mapping, Management of Metadata, and efficient Visualization of Data Samples. But most of the existing models that perform these tasks either have limited scalability or showcase lower processing efficiency on the collected data samples. To overcome these issues, this text proposes design of an efficient multidomain augmented data aggregation model to solve heterogeneity issues for IoT deployments. The proposed model initially converts all input data samples into 1D vectors via convolutional flatting operations. These vectors are further converted into multidomain feature sets via a combination of Frequency, Gabor, Entropy, Wavelet and Convolutional analysis. These feature sets assist in identification of differential & spatial data patterns, which can be used for further analysis. To demonstrate efficiency of the proposed model, the collected IoT datasets were given to an Auto Regressive Integrated Moving Average (ARIMA) Model for prediction of temperature and humidity levels. These predicted levels were compared with existing models, and it was observed that the proposed model was able to improve the accuracy of prediction by 8.3%, while improving the precision by 5.9%, and recall by 2.5% under real-time deployment data samples.