The Anonymization Protection Algorithm Based on Fuzzy Clustering for the Ego of Data in the Internet of Things

Algorithm 3

Anonymization protection of the ego of data based on fuzzy clustering in the Internet of things.

 Input: , , Spactab, Sampdata;

 Output: processed data set;

 Begin

 (1) Do Algorithm 1; % All nodes are implemented by fuzzy clustering according to the

   spatial location and get the node clustering set Subndgrp.

 (2) FOR each in Subndgrp % Get a subset of nodes from Subndgrp

   orderly.

 (3)    Do Algorithm 2; % The sampling time fuzzy clustering is performed on the data

      records of the nodes in the subset, and the equivalence class set is obtained. Anonymization

      processing for each equivalence class.

 (4)     FOR each in Substgrp % Get a subset of nodes

       from Substgrp orderly.

 (5)     Psum = 0; % Psum denotes the total number of nodes in an equivalence class.

 (6)        FOR each in

 (7)         sum = sum + ; %   represents the -axis of node in the

           Spactab.

 (8)           sum = sum + ; %   represents the -axis of node in the

           Spactab.

 (9)           sum = sum + ; %   represents the -axis of node in the

           Spactab.

 (10)       ENDFOR

 (11)     temp = sum/psum; temp = sum/psum; temp = sum/psum;

 (12)     FOR each in Spactab % Replace the spatial location

         information of the node number in the equivalent class data record with the spatial attribute

         NODESP (, lambda) of the equivalent node of the equivalence class.

(13)            IF then

(14)             ; ; ;

(15)           ENDIF

(16) ENDFOR

(17) = mid (·Sampling time,…, ·Sampling time) % Equivalent sampling

        calculated with the intermediate value of the sampling time of all data records of

        time is equivalence class.

(18) For each in

(19) % Replace the sampling time attribute of

          the record with the sampling time in the equivalent class.

(20) ENEFOR

(21) ENDFOR

(22) ENDFOR

(23) Count the rest of the data to delete.

 % A small number of records cannot be clustered, because the special distance or the

duration of sampling time between those records and the most number of records. If the few

records are putted into the equivalent class, the nodes’ sampling duration of the equivalent class

is greater than , or the spatial contiguity of nodes in the equivalent class goes beyond . %

 END