A Temperature Data Denoising Method with Consideration of Noise Variance

Abstract

In this paper, a temperature data denoising method is presented by considering the noise variance. First, conventional smoothness-based denoising method is briefly reviewed. Then, the minimization of closeness function of conventional method is replaced by the minimization of negative log likelihood function to develop a denoising method in which probability density function of noise is assumed to be a zero-mean Gaussian function with different variances. The optimal denoised temperature data is easily obtained by solving matrix inversion if the noise variance is known in advance. Finally, the temperature data collected from the sensor networks in USA and Taiwan are used to show the effectiveness of the proposed denoising method.