A lightweight data transmission reduction method based on a dual prediction technique for sensor networks

An essential design concern in a resource-constraint sensor network is optimizing data transmission for each sensor node (SN) to prolong the network lifetime. Many research works cited that the dual prediction technique remains the most efficient technique for data reduction. A large amount of redundant data is usually transmitted across the network, leading to collisions, loss of data, and energy dissipation. This article proposes a data transmission reduction method (DTRM) to solve these problems, implemented on the cluster heads and operates in rounds. DTRM is lightweight in processing, has low complexity costs, and needs a limited memory footprint, but it is robust and effective. It can be combined with any form of cluster-based data aggregation. We have incorporated the proposed DTRM with the data aggregation-adaptive frame method (DA-AFM), implemented on the SNs within the clusters. DA-AFM can eliminate temporal redundancies and correlations in the sensor's time-series readings. This helps the SN take fewer readings, which improves the efficiency of reducing data transmission and decreases the amount of energy spent during sensing. The proposed DTRM approach decreases the average transmission rates of data while maintaining data quality. This study is evaluated on real data obtained from the Intel Berkeley Lab and compared with three recent data reduction techniques focused on prediction. The results show that DTRM consumes up to 70% less energy while preserving the expected quality of data and reducing transmission.