Network Host Cardinality Estimation Based on Artificial Neural Network

Algorithm 2

Generate training data.

(i)	Input: SIP，LEA， g//The length of LE
(ii)	H1//hash function used in LE, used to map a host to a bit in LE
(iii)	Output: Train data set: trainX, predY′
(1)	For each {sip, RC, OIP} in SIP
(2)	LEset = {le1, le2,…,} ← Locate LE related to sip in LEA；
(3)	ULE ← Unionize every LE in LEset by bitwise “AND” operation；
(4)	For each bip in OIP
(5)	i ← H1(bip)
(6)	Set the ith bit in ULE to 1
(7)	Endfor
(8)	EC ← calculate the estimating cardinality of aip’ according to ULE
(9)	Insert [EC, ULE] into trainX
(10)	Insert (RC-EC) into predY′
(11)	Endfor
(12)	Return {trainX, predY}