Table - PMC

Skip to main content

View full-text article in PMC

. 2015 Oct 30;15(11):27760–27782. doi: 10.3390/s151127760

Algorithm2 Clustering algorithm

(1)
initialization:
- (a)
  calculate the distances between the PN $s_{P N}$ and the CNi $s_{i}$ as $d_{0}^{i} = | {\vec{s}}_{i} - {\vec{s}}_{P N} |$ and the distances between the CNi and the fusion center $s_{F C}$ as $d_{1}^{i} = | {\vec{s}}_{i} - {\vec{s}}_{F C} |$ for i = 1, 2, …, D;
- (b)
  choose 2K CNs with the shortest $d_{0}^{i}$ and select K reference nodes with the shortest $d_{1}^{i}$ from the 2K CNs as the initial cluster heads $v_{k}$ for k = 1, 2, …, K;
- (c)
  set the initial clusters $C_{k} = {v_{k}}$ and the initial cluster centers $c_{k} = v_{k}$ for k = 1, 2, …, K.
(2)
repeat:
- (a)
  allocate CNi for i = 1, 2, …, D − K to the nearest cluster k_i where $k_{i} = \underset{1 \leq k \leq K}{\arg \min} | {\vec{s}}_{i} - {\vec{c}}_{k} |$ and let $C_{k_{i}} = C_{k_{i}} \cup {s_{i}}$ ;
- (b)
  to each cluster, update ${\vec{c}}_{k}$ by averaging all the ${\vec{s}}_{i}$ of $s_{i} \in C_{k}$ ;
- (c)
  reselect the cluster head of each cluster by $v_{k} = \underset{s_{i} \in C_{k}}{\arg \min} | {\vec{s}}_{i} - {\vec{c}}_{k} |$ ;
- (d)
  reinitialize $C_{k} = {v_{k}}$ and $c_{k} = v_{k}$ for k = 1, 2, .., K;
until: all the CNs in each cluster are not changed.
(3)
Output: $C_{k}$ for k = 1, 2, .., K.