TY - JOUR
T1 - How NAT-compatible are VoIP applications?
AU - Lin, Ying-Dar
AU - Tseng, Chien-Chao
AU - Ho, Cheng Yuan
AU - Wu, Yu Hsien
PY - 2010/12/1
Y1 - 2010/12/1
N2 - Today, the Internet is very diversified, further complicating the classic NAT traversal problems. In order to solve these problems, there are many proposed methods classified into two categories. One enhances the NAT traversal techniques of applications, and the other tries to modify the behavior of NATs. In this article we focus on the former because NATs have been installed, and their behavior cannot be altered through endpoint users. Accordingly, in order to test NAT traversal techniques of five VoIP applications (Skype, MSN, Google Talk, X-Lite, and Linphone), three network topologies have been designed with two endpoints behind the same, different, or multilevel NATs. Through a series of experiments and from the experiment results, we observe that these VoIP applications use some traversal techniques, such as NAT mappedaddress probe, peer discovery, path check, and relay first, proposed by STUN, TURN, and ICE to make a direct connection when endpoints are behind the same or different NATs with independent mapping rules. However, with multilevel NATs, no endpoints can establish a direct connection when they use the above mentioned techniques, even if hairpin behavior is supported by NATs.
AB - Today, the Internet is very diversified, further complicating the classic NAT traversal problems. In order to solve these problems, there are many proposed methods classified into two categories. One enhances the NAT traversal techniques of applications, and the other tries to modify the behavior of NATs. In this article we focus on the former because NATs have been installed, and their behavior cannot be altered through endpoint users. Accordingly, in order to test NAT traversal techniques of five VoIP applications (Skype, MSN, Google Talk, X-Lite, and Linphone), three network topologies have been designed with two endpoints behind the same, different, or multilevel NATs. Through a series of experiments and from the experiment results, we observe that these VoIP applications use some traversal techniques, such as NAT mappedaddress probe, peer discovery, path check, and relay first, proposed by STUN, TURN, and ICE to make a direct connection when endpoints are behind the same or different NATs with independent mapping rules. However, with multilevel NATs, no endpoints can establish a direct connection when they use the above mentioned techniques, even if hairpin behavior is supported by NATs.
UR - http://www.scopus.com/inward/record.url?scp=78650870294&partnerID=8YFLogxK
U2 - 10.1109/MCOM.2010.5673073
DO - 10.1109/MCOM.2010.5673073
M3 - Article
AN - SCOPUS:78650870294
SN - 0163-6804
VL - 48
SP - 58
EP - 65
JO - IEEE Communications Magazine
JF - IEEE Communications Magazine
IS - 12
M1 - 5673073
ER -