A Peak Prediction Method for Subflow in Hybrid Data Flow

Algorithm 1

Choose flow.

	Input: the list of steams; the size of window; the number of predicted steam
	Output: the number of subsidiary steam; the number of delay time
(1)	procedure chooseSteam ()
(2)	for iterate through the list of steams do
(3)	for iterate through all of the number of delay time do
(4)	for iterate through all of the size of window do
(5)	Calculate and get the correlation coefficient between the delay time and window
(6)	Summing up the correlation coefficients
(7)	Calculate and get the mean of the correlation coefficient
(8)	Update the maximum delay correlation coefficient and the delay time
(9)	Update the maximum delay correlation coefficient and the delay time
(10)	Update the number of the auxiliary data flow
(11)	return the number of the auxiliary data flow and the delay time