Sequence Alignment with Dynamic Divisor Generation for Keystroke Dynamics Based User Authentication

Algorithm 1

The pseudocode to explain the sequence alignment algorithm.

Input: Training data extracted from a genuine user, . Testing data extracted from a genuine user or an imposter, .

In the training phase:

(1) for each attribute   do

(2)   ←

(3)   ←

(4) end  for

(5) for each attribute   do

(6)  diff ←

(7)  rangeDiff← diff/20

(8) end  for

(9) for each row and each attribute in   do

(10) label ← ceil(()/rangeDiff)

(11)  if label = 0

(12)  label ← 1 // If the data is exactly the same value as minimum time, then it should categorize as label

(13) model← label // We just keep it as to represent

(14) end for

In the testing phase:

(1) Convert the to alphabet letter format based on the minimum point and the range difference from step 3 and step 7

   respectively, from training phase, and then we just run one time from step 9 to step 14 from training phase.

(2) for each row in model do

(3)   for each attribute in   do

(4)    if = model

(5)    Score← 1

(6)   end for

(7)  Final_Score← sum(Score)

(8) end for

(9) Checking_Score ← mean(Final_Score) // Can be max, min, median, mean, and mode

(10) return Checking_Score

Output: The score then used to compare the threshold