Algorithm 3.

MDAV′(D)

Require: Original dataset D, privacy parameter k, weights wICD   and wAge
Return: Anonymized dataset D̃, ILM(D̃) and ALM(D̃)
1:	D̃ ← Ø; ĩ ← 0; ã ← 0; p ← ∑T∈D |T|
2:	while |D| ≥ 3* k do
3:	F ← the most frequent trajectory in D  ▷Find the most distant trajectory to F
4:	X ← argmaxT∈D{(i + a)|i, a ∈ A-GS(F, 0, 0, T)}
5:	{C, i′, a′} ← formCluster(X, k)
6:	D̃ ← D̃ ∪ C; ĩ ← ĩ + i′; ã ← ã + a′
7:	Y ← argmaxT∈D{(i + a)|i, a ∈ A-GS(X, 0, 0, T)}
8:	{C, i′, a′} ← formCluster(Y, k)
9:	D̃ ← D̃ ∪ C; ĩ ← ĩ + i′; ã ← ã + a′
10:	end while
11:	while |D| ≥ 2 * k do
12:	F ← the most frequent trajectory in D
13:	X ← argmaxT∈D{(i + a)|i, a ∈ A-GS(F, 0,0, T)}
14:	{C, i′, a′} ← formCluster(X, k)
15:	D̃ ← D̃ ∪ C; ĩ ← ĩ + i′; ã ← ã + a′
16:	end while
17:	R ← select a trajectory from D uniformly at random
18:	{C, i′, a′} ← formCluster(R, |D|)
19:	D̃ ← D̃ ∪ C; ĩ ← ĩ + i′; ã ← ã + a′
20:	return {D̃, ĩ/(p * wICD), ã/(p*wAge)}