Technology and Research
Intel® Technology Journal Home
Volume 10, Issue 01
Converged Communications
Table of Contents
Technical Reviewers
About This Journal
Intel Published Articles
Read Past Journals
Subscribe
E-Mail this Journal to a Colleague
Main Visual Description
Intel Technology Journal - Featuring Intel's Recent Research and Development
Converged Communications
Volume 10    Issue 01    Published February 15, 2006
ISSN 1535-864X    DOI: 10.1535/itj.1001.04

  Section 3 of 11  
Quality Campus VoIP: An Intel® Case Study
VOICE INFRASTRUCTURE

In this section we provide an overview of the technical voice infrastructure with an emphasis on characteristics important for voice quality and QoS, which include connecting the IP PBX with the PSTN, gateway integration, and a general discussion of the IP telephony endpoints.

The IP PBX resides behind the traditional PBX, meaning that no changes were needed to provide the connection between the premises and the public telephone network. All calls continue to come into and out of the campus through the existing PBX. The PBX continues to route inbound calls as usual, directly or through the auto-attendant without modification. This configuration enabled quick installation of the new system and removal of the old, with minimal disruption.

To make the connection between the traditional PBX and the IP PBX, a gateway is used to enable translation of TDM circuit-switched calls from the existing PBX to packet-switched calls from the new IP PBX. In this case the direct physical connection to the PBX is through a PRI ISDN T1 interface. The gateway is connected to the LAN through a switch, as shown in Figure 1, making sure that the configured protocols are matched up properly.

Pilot users tested a variety of IP telephony endpoints, including existing legacy phones and IP phones. Most users received IP phones. Legacy phones are connected to the IP PBX using Intel's PIMG as shown in Figure 2. IP phones are connected directly to the LAN.



Figure 1: Connecting the IP PBX to the traditional PBX
click image for larger view
 



Figure 2: Using PIMG to connect legacy phones
click image for larger view
 

We tested several types of IP-based phones in the pilot, including hardware phones, software phones, and wireless phones. All these were SIP-enabled out of the box. Of users with IP phones, 58% received a hardphone, 28% received a softphone, and 14% received both. Wireless IP phones were tested but not deployed to any live users.

Some hardware IP phones were deployed with Power over Ethernet (PoE). PoE makes for a more streamlined approach since a separate AC adapter for each phone is expensive and takes up valuable outlet space. PoE enables the transmission of electrical power over the same Ethernet cable as voice and data, using industry- standard IEEE 802.3af. For implementation, the access switch had to support the PoE feature, so in some instances the switch was replaced in this trial. An advantage of PoE is that it offers backup power capabilities if power goes out in the facility. Since the PoE cable draws power from the network, power during an outage continues to be supplied via the network's Uninterruptible Power Supply (UPS). This improves the user experience and overall perception of voice quality.

Another important feature of the hardware IP phones with respect to QoS is the ability to do 802.1q virtual LAN (VLAN) trunking and the ability to mark outgoing packets using differentiated services or DiffServ. In areas where these phones were on a separate VLAN, the default VLAN was used. In others, 802.1q and DiffServ were used.

Software phones or softphones consist of a software client installed on a PC, connected to a headset with audio and microphone. Good quality audio devices are essential for a good end user experience. The softphones, like hardware IP phones, have the ability to mark the packet with DiffServ allowing those packets to obtain priority on a properly configured network. However, there were other factors that also impacted voice quality. For example, Intel tested wired and wireless headsets from a variety of vendors and found that the selection had a profound impact on call quality and usability.

  • Wireless headsets–Several types of Bluetooth* wireless headsets were tested and none were able to provide acceptable call quality. Noise and echo were problematic. Further, there were major usability issues concerning power including battery life and awkwardness of turning power on to answer an incoming call.
  • Wired headsets–Wired headsets come either with their own Digital Signal Processor (DSP) using the PC's USB port, or they use the PC soundcard through the headset and microphone plugs. Users experienced a wide range of quality issues including noise and echo. The more expensive USB headsets with DSP provided the best quality.

  Section 3 of 11  

In This Article
Abstract
Introduction
Voice Infrastructure
Voice Quality Plan
LAN Design
Trial Voice Quality Results
Key Challenges and Solutions
Conclusion
Acknowledgments
References
Authors' Biographies
Download a PDF of this article.   
Email This Page
Back to Top