Enhancing the Computational Intelligence of Smart Fog Gateway with Boundary-Constrained Dynamic Time Warping Based Imputation and Data Reduction

Internet of Things (IoT) is one of the fast-growing technology that exhibits are markable attention among the various fields with a wide range of applications. Owing to the rapid explosion of the IoT application usage, the production rate of the streaming data has increased, which tends to the difficulty of acquiring knowledge from the massive amount of data. Hence, it is essential to apply the preprocessing technique on the raw data in the smart gateway to remove the noisy, irrelevant, and missing data instead of processing the entire data on the Cloud. However, performing the imputation and relevant data extraction of stream data within a short period of time is a challenging task. To resolve this constraint, this paper presents the accurate Data Imputation for Satisfying all the MIssing values and the Separation of data points from the Stream data (DISMISS). The DISMISS approach focuses on handling the missing values and extracting the relevant information from the stream data in the most effective way. To improve the time efficiency of the stream data processing, the DISMISS approach applies the boundary Binning method in the gateway, which reduces the noise in the raw stream data. In consequence, it accurately imputes the missing values in the noiseless incomplete stream data using the Boundary-constrained DTW method. Finally, the DISMISS approach selects the most representative points by applying the Perceptually Important Points (PIP) method along with the Particle Swarm Optimization (PSO). This optimal PIP technique with PSO effectively extracts only the valuable data points from the entire data stream. Thus, this approach effectively evades the unsolicited data from the raw stream data in the smart gateway itself and then, offloads the reduced set of valuable data alone to the cloud. Thus, the implementation results prove the significant outcome of the DISMISS approach outperforms the existing method in terms of the imputation of missing values.