|
VOICE From the January 2006 issue of Communications News |
Ensure QoS in your VoIP VoIP consists of numerous components and moving parts, as well as many protocols, which may result in a loss of voice communication. In addition, because there are multiple vendors creating such parts, compatibility becomes an issue. Even when successful communication is achieved, voice quality must then be ensured. These issues need to be detected and isolated for better VoIP management. Outages. Any one of the above VoIP elements can fail at any given time, and detecting an outage can become cumbersome and time consuming–and difficult to isolate the problem in a satisfactory time frame. An event-driven alarm system detecting vital problems is one solution. By implementing an intelligent (management) server that is ready to receive alarms and send relevant information through elements like a SNMP trap, e-mail/pager, carrier console, or an intelligent console, the finger pointing should diminish, as should the network managers’ workload. The VoIP management server could also check vital signs by polling the connections. To determine whether a specific IP address is accessible, a ping sends a packet, or message, to another computer and waits for acknowledgement. The nature of acknowledgment received determines the IP address’ availability. Blind polling sends a ping to every element in the network a predetermined number of times per minute. This type of polling monitors the entire network at the expense of bandwidth. Therefore, a better polling option is severity-based. Severity-based polling transmits a ping to critical elements more often. Dynamic polling could also be introduced to baby-sit any failed pieces by increasing the number of pings to that particular piece. The number of pings would be recalculated based on the number of previous faults. Voice quality. Since voice shares the same network with data, care should be taken to ensure that voice quality is preserved. Users demand that the quality of VoIP match the business quality that PSTN voice routinely provides. Perceived voice quality is a function of many factors, including delay, line noise and echo, and jitter. Each listener determines voice quality differently. As a result, a mean opinion score (MOS) was created by the International Telecommunication Union, Telecommunication Standardization Sector (ITU-T), as a benchmark to determine the quality of sound. A wide range of men and women rated the quality of preselected voice samples on a scale from 1 (bad) to 5 (excellent). The scores were averaged to provide the MOS for each sample. Communication quality is rated in the 3-4 range. Given that the MOS is a highly subjective interpretation, the ITU-T created an automated scoring process, with the latest standard for assessing voice quality being the perceptual evaluation of speech quality (PESQ). PESQ adds additional processing steps to account for signal-level differences and the identification of errors associated with packet loss. A total quality management and measurement tool is needed to not only detect but also isolate any voice quality problems. Both real-time and trend information are essential to management views and reports. A central server can be created to dynamically collect performance/delay data from various sources to continuously calculate variable polling periods and inject active packets, as necessary. The server can analyze the importance of each problem and report them to a console via an alarm. The console would send instructions back to the server to inject appropriate active test packets into the VoIP network to pinpoint the trouble spots. The console can then display overall call quality with display drill down to detail elements with MOS. This article was provided by Choon B. Shim, chief
technology officer and senior VP of engineering of Qovia, Frederick, Md.
|