Energy Conserving Communication in WSN based on static data prediction by using SEKF

Abstract

Wireless Sensors Networks (WSNs) recently, have drawn a lot of attention.  Despite its potential applications in a wide range of fields, wireless sensor nodes' restricted communication bandwidth, inadequate processing units, little memory, and power limitations severely limit their capabilities. One of the main challenges in this area is extending the life of battery-powered sensors in WSNs by reducing energy usage. This problem is addressed using a variety of techniques, such as deep learning, machine learning techniques, statistical techniques, and time series forecasting.  One strategy is to utilize data prediction to reduce the volume of transmitted data without sacrificing its quality. This paper presents a model for wireless sensor networks energy saving using the Static Extended Kalman Filter (SEKF). The technique is used to accurately dual predict. The plan consists of two stages. In the first stage, the transmission from the sensor node to the sink node is reduced based on four steps (data equality, data deviation computation, faulty data detection, data reduction based on prediction). In the second stage, the data is reconstructed at the sink node to maintain system reliability. The proposed model demonstrated superior performance compared to other methods, reducing data throughput in the first phase by 60.72%. In the second phase, data was reconstructed with 97.86% accuracy at a data reduction rate of 62–60%, with an energy consumption of 3.928 J. These results were achieved by SEKE for single-node reconstruction. Furthermore, the proposed model performed well when applied to data containing negative values, achieving acceptable data reduction with accuracy ranging from (94-95%) in several experiments. The Intel Berkeley Research Lab (IBRL) dataset was used for all experiments.