Complexity Analysis of Big Data Utilizing Lifting Based DWT for Multimedia Sensor Networks

Abstract

Big data such as images and videos captured by camera sensors possess large volumes of data and also occupy large memory, but the Wireless Sensor networks have lower memory and lower computational resources to store the big data. Due to this limitation, it is not suitable to preprocess the data which are collected by sensor networks for energy and bandwidth efficient transmission over sensor networks. The modern signal processing techniques such as Discrete Wavelet Transform (DWT) based image coding is computationally complex, so recently, for low-processing power nodes, the Lifting-Based wavelet filters are proposed to be used as a low-complexity wavelet transform schemes. So the proposed research work presented in this paper compares the computational complexity and performance of Haar, 5/3 and Daubechies 9/7 filter based DWT on images using the Lifting Scheme and Conventional Scheme. Later from the results, it is analysed that the computational complexity of the lifting scheme comes out to be almost half of the conventional scheme of computing DWT on images.