Privacy-Preserving Publication of Time-Series Data in Smart Grid

Algorithm 5

Clamping wavelet perturbation algorithm (CWPA).

Inputs:
(i)	Consumptions: with for all
(ii)
(iii)	The maximum magnitudes of first DWT coefficients:
(iv)	Privacy budget:
(1)	Clamping: for each individual time-series consumption ,
(i)	compute the first DWT coefficients of :
(ii)	if , then replace with for
(2)	Laplacian Mechanism: compute the sum of noisy consumptions of each DWT coefficient: for all . We denote
(3)	Pad with zeroes; the result is denoted by
(4)	Compute the inverse DWT of to get the noisy sum of consumptions denoted by of the initial sum