An improved B2BUAWM approach for VoIP
Infrastructure
Fernando Barreto
Coordenation and Management of Information Technology
Federal University of Technology – Parana (UTFPR)
Apucarana, Brazil
fbarreto@utfpr.edu.br
Abstract—This article presents a study for VoIP technology to
elaborate a differentiated VoIP infrastructure with the proposed
approach ProxySIP Bridge. This approach enhances the Back-to-
Back User Agent With Media (B2BUAWM) approach in order to
isolate the voice traffic from the remaining general purpose
traffic and to enable direct voice traffic whenever possible. A
differentiated VoIP infrastructure with ProxySIP Bridge is pre-
sented, which is an alternative to B2BUAWM original approach.
Keywords-component; VoIP infrastructure, B2BUAWM,
ProxySIP Bridge
I. INTRODUCTION
With the advances of network computer technologies, the
IP network infrastructure can provide a voice over IP (VoIP)
service at an acceptable level of quality. The VoIP service can
significantly reduce the phone call cost when compared to the
standard public switch telephony network. This technology is
helpful to companies composed of many affiliates, who
recognize it as a chance to significantly reduce telephone costs.
In the academic environment, this reduction is also interesting,
mainly if a university is composed of various campuses.
There are many important points that should be considered
in a VoIP infrastructure such as Firewall, Network Address
Translation (NAT) and Quality of Service (QoS). Generally,
these points make the VoIP infrastructure very difficult to
implement. In order to achieve a facilitated VoIP infrastructure,
the B2BUAWM approach can solve Firewall and NAT
problems, because it needs only small modifications on
Firewall/NAT and no support from VoIP software client.
However, B2BUAWM presents some problems of overhead
which can be solved with an enhanced B2BUAWM approach
named ProxySIP Bridge. The ProxySIP Bridge allows a
differentiated VoIP infrastructure which deviates the voice
traffic from Firewall/NAT, reduces the number of hops needed
to forward the voice traffic between the end users equipment
and, whenever possible, can enable voice traffic to occur
directly between them. The ProxySIP Bridge approach was
implemented using open source software, which enables its
adaptation by the academic community.
II. VOICE OVER IP
The establishment of a VoIP call needs a signaling
protocol. This signaling is made by Session Initiation Protocol
(SIP) [4]. In order to transport the data stream containing the
digital audio voice the SIP use the Real-Time Transport
Protocol (RTP) [6] and, to monitor the RTP traffic, they use
the RTP Control Protocol (RTCP) [6].
A. SIP
The SIP protocol is specified in [4]. A hardware equipment
that acts on behalf of a user and implements the SIP protocol is
denominated User Agent (UA). The main methods defined to
establish and finish a VoIP call are: INVITE, ACK and BYE.
The INVITE and BYE methods need an answer from remote
UA to confirm the method operation. The INVITE method
carries SIP information to identify the source UA, the
destination UA and the Session Description Protocol (SDP)
[5], which carries additional information from source UA, like
Codec details, network IP address and port number for RTP
session. When the destination UA accepts the INVITE message,
it sends a message with an OK code containing a similar SDP
with information for RTP session. When the source UA
receives the OK message, it sends an ACK message to the
destination UA to confirm the RTP session and begins the
voice transmission. When the destination UA receives the
ACK, it also begins the voice transmission. Further details on
SIP message flows and methods are presented in [4].
In a VoIP infrastructure with many UAs that belong to the
same SIP domain, a centralized Proxy service [4] is
recommended for SIP domain, here called ProxySIP. The
objective is to facilitate the UA localization inside a SIP
domain. The ProxySIP controls, authenticates the UAs and is
also responsible for routing the SIP messages. This enables the
management of SIP messages from INVITE to BYE (SIP
transactions and SIP dialog) [4].
B. Codec and Quality of Service Concepts
The voice RTP session is configured according to the
coder/decoder (Codec) information exchanged through SDP.
More details about Codecs can be found in [14]. The quality of
a Codec can be measured by the Mean Opinion Score (MOS)
index. Table I presents a few standard Codecs. The Voice
Bandwidth is the constant bit rate needed for Codec operation.
The Total Bandwidth takes into consideration the overhead of
all TCP/IP headers in an Ethernet network (78 bytes) and also a
978-1-4577-1792-5/11/$26.00 ©2011 IEEE