Editor's note: Part three of our SIP Trunking Explained series looks at how much VoIP bandwidth you'll need to ensure high-quality SIP trunking services -- whether you're connecting to a SIP trunking provider via a dedicated data line or an Internet connection. Check out the rest of the series (see box below) for essential information on SIP vs. PRI, selecting a SIP trunking provider, how to enable your legacy equipment, the advantages of SIP trunking, and security precautions to reduce toll fraud.
When SIP trunking is implemented via a dedicated data line to the SIP trunking provider, the necessary bandwidth is calculated by the provider according to the number of simultaneous streams (calls) and Voice over IP (VoIP) call quality required for each stream.
When we talk about VoIP call quality, think of the term codec, which stands for coder-decoder. Here's the basic explanation of how VoIP works: A codec converts an audio signal (voice) into compressed digital form for transmission, then converts it back into an uncompressed audio signal for replay.
The most popular and frequently supported codecs are:
- G.711: The standard that digitizes analog voice signals and requires 64 Kbps of bandwidth
- G.729: The standard that digitizes analog voice signals using an algorithm for an 8:1 compression rate using 8 Kbps
- G.722: The standard that provides higher-fidelity speech and uses 48 to 64 Kbps
- G.726: The standard that requires 16 to 40 Kbps of bandwidth and is primarily used in international trunks in the phone network.
Most service providers work with G.711 to ensure good voice quality when the SIP trunk is provided via a dedicated data line. Note that the bandwidth requirements above do not include IP overhead, which is an additional 23 Kbps. This means a G.711 codec requires 87 Kbps of bandwidth per call. G.711 provides the same quality as a landline-to-landline telephone call, while G.729, using 8:1 compression, provides quality similar to that of a mobile call.
If the SIP trunk connects to the provider via the Internet, call quality can be seriously affected when there is heavy Internet traffic, causing VoIP packets to be delayed, dropped or lost during a call. In these cases, I recommend using a dedicated Internet connection so VoIP calls are routed through it with minimal delay.
Latency is one of VoIP's biggest enemies. Latency is measured in milliseconds and is calculated in terms of round-trip delay from source to destination and back. Callers will begin to notice round-trip voice delay of about 250 to 300 ms or greater, which is why latency should be kept under 150 ms one way (source to destination). The further away the SIP provider, the bigger the latency will be.
On average, an IP packet travelling from Europe to the U.S. and back takes about 120 to 150 ms. The latency of an IP packet travelling within the U.S. will be much lower than that of a packet travelling to Europe and back, due to the shorter distance.
It's always a good idea to test the latency to your SIP provider by simply pinging their SIP Trunk endpoint. This is a very simple and revealing test that will help you get an indication as to whether latency will be an issue with the provider.
Some companies terminate SIP trunks to their branch offices. Instead of renting a Primary Rate Interface (PRI) line, which would be very expensive because of the branch's location, companies send the SIP trunks through their wide-area network infrastructure, utilizing the existing network and avoiding additional telecommunications costs. Again, it is necessary to ensure the round-trip delay is kept below 150 ms.
Next: In part 4 of SIP Trunking Explained, find out why SIP trunks are becoming the preferred method of providing VoIP communications in the enterprise.
How to use SIP trunking services in your WAN strategy
Learn how to prioritize your VoIP traffic to address delay
Over-the-top providers get the call for SIP services
What do you know? Take the VoIP quiz and see