How we did it
These voice-over-IP gateways were tested according to a comprehensive methodology created by Mier Communications, first applied in 1997, and continually refined and enhanced since. The test-bed topology varies somewhat depending on the metric being tested or measured. An assortment of test systems and tools are applied at one point or another.
For this particular review, these voice-over-IP gateways were configured on separate "remote" LAN subnets that were separated by an IP WAN consisting of Cisco 4700 and 7200 LAN/WAN routers. For testing voice-over-IP voice quality and interactive call quality, each LAN subnet was connected to the IP backbone via a Cisco 4700 router and T-1 WAN links. The T-1 passed through an Adtech SX/12 T-1-WAN simulator, which introduced errors, latency and jitter as specified by the tester. Other WAN transmission environments and anomalies were introduced via the Shunra Cloud, a Windows NT-based software simulation environment from Shunra Software.
To test load (up to four full T-1s of bidirectional call load are applied in these tests) and call-completion rate, the LAN subnets were connected to the IP backbone via Cisco 7200 routers and T-3 WAN links. The plain old telephone system part of the test bed consisted of TSU 100 multiplexers and Atlas 800 switching equipment from Adtran.
Hammer Technologies' Hammer IT Call Center Telephony Load Testing System and Hammer Loadblaster 500 systems were used to apply up to four full T-1s of call load, and to send and record the male and female voice recordings that were then rated by blind panels to yield the voice quality mean opinion score (MOS) ratings. The Hammer equipment was also used to measure latency.
The Telegra FAX test system from Agilent Technologies was used in this test round to test fax transmission through the voice-over-IP gateways. Conventional PCs with V.90 modems were used to test modem transmission though the voice-over-IP network. The Agilent Internet Advisor, a network analyzer, was used to measure the packet size, packet rate and bandwidth of each different voice-over-IP setting.
Because blind panel MOS voice-quality ratings are based on voice recordings that are sent one-way through the voice-over-IP network, they do not address or measure some aspects of voice over IP that also greatly affect call quality, such as latency and bidirectionality. Mier Communications has developed its own tests and procedures for objectively assessing interactive call quality, which focus specifically on these other parameters affecting overall voice-over-IP quality. These tests and procedures were applied for this review.
RELATED LINKS
Interactive scorecard and NetResults
