A Lightweight Intrusion Detection Method Based on Fuzzy Clustering Algorithm for Wireless Sensor Networks

Algorithm 1

Training Stage.

Input: Training set X=,q

Output: Normal clustering, noise clustering and one-class SVM model

(1)  use feature A of X

(2)  lc = storage local cluster center

(3)  gnorc = 0/global normal cluster center

(4)  gnosc = 0/ global noise clustering center

(5)  k = 0

(6)  calculate d = / calculate the distance between any two points

(7)   = cutoffdis(q)/ cutoff distance, equation (9)

(8)   / equation (9)

(9)   for  i = 1 to n  do

(10)find in the field of /the field is a circle with radius and center

(11)if   =   then

(12)lc.append()/   is a local cluster center and stored

(13)k = k+1

(14)end if

(15)  end for

(16)  for  i = 1 to k  do

(17)if   is min  then/   in lc

(18)gnorc =

(19)end if

(20)if   is max  then/   in lc

(21)gnosc =

(22)end if

(23)  end for

(24) initialize FCM with gnorc and gnosc

(25) X is divided into normal clustering data and noise clustering data by the FCM

(26) use feature B of normal clustering data

(27) input feature B of normal clustering data to SVMDS

(28) train one-class SVM by SVMDS

(29)  return normal clustering, noise clustering and one-class SVM model