Preserving Differential Privacy for Similarity Measurement in Smart Environments

Algorithm 3

Anonymizer’s computation phases.

Computation phases for Anonymizer  𝓐

Input: Computation request ℛ

Output:ℱ^′_𝒮𝒞(𝑋,𝑌)

/*Phase 4*/

Initialise: 𝑎=0, 𝑏=0, 𝑐=0 and 𝑑=0;

Receives Enc_pk𝒞(𝓉) from Alice (from Phase 1);

Receives 𝑀^′ from Bob (from Phase 3);

// decryption operation to obtain 𝓉

Computes Dec_pr𝒞(Enc_pk𝒞(𝓉));

for  𝑖=1  to  𝑛  do

   // decryption operation to obtain 𝑥^′_𝑖𝑦^′_𝑖

   Computes Dec_pr𝒞(Enc_pk𝒞(𝑥^′_𝑖𝑦^′_𝑖));

  if  𝑥^′_𝑖𝑦^′_𝑖=𝓉³  then

   Increases 𝑎 by 1;

  else if  𝑥^′_𝑖𝑦^′_𝑖=𝓉²  then

   Increases 𝑏 by 1;

  else if  𝑥^′_𝑖𝑦^′_𝑖=𝓉¹  then

   Increases 𝑐 by 1;

  else if  𝑥^′_𝑖𝑦^′_𝑖=𝓉⁰  then

   Increases 𝑑 by 1;

  else

    stop (error)

  end if

end for

/*Phase  5*/

for each ℛ  do

  Computes ℱ_𝒮𝒞(𝑋,𝑌);

  Generates Laplacian noise Lap(Δ𝐹/𝜖);

  Returns ℱ^′_𝒮𝒞(𝑋,𝑌)=ℱ_𝒮𝒞(𝑋,𝑌)+Lap(Δ𝐹/𝜖);

end for