So you’ve built your VoLTE network for high quality. You are providing high bandwidth to your customers so browsing and watching video is great. You also managed to get delay, jitter and packet loss down so voice calls are using WB-AMR giving your customers top quality.
But even though VoLTE is set to succeed, it will not happen overnight and it can’t be an island, there are other networks it needs to connect with. You need to allow for calls to GSM, PSTN and wireline VoIP.
To allow for this, you engineered your network in a way that if a call needs to exit the VoLTE network to any of the other networks you simply transcoded it to G.711. With that, it connects “without any issue” to any network out there.
So you connected your VoLTE network to other networks but that’s exactly when problems started to come up. Users started calling your contact center complaining about voice quality issues, they felt as if a mouse has cut the wire. Analyzing the complaints showed that there are problems not only when calling between the different types of networks but even when calling between VoLTE networks of different service providers.
Different type of networks perform differently
Think of a VoLTE network. It prioritizes voice over other data types. It will not retransmit voice packets even if there is packet loss (different from its handling of data) and it dynamically tunes rate, stuffing multiple voice samples into a single packet and add redundancy when needed.
In a VoLTE to VoLTE case, this results in great quality but when the destination is on a different network, that network handles things differently, resulting in quality issues.
On the GSM network, data will be retransmitted in case of packet loss, VoIP traffic included. This results in increased delay. Moreover, the 3G network works in burst mode sending a large amount of data to the core of the network in bursts, resulting in delay and jitter.
In case of voice TDM calls, NB-AMR is used which is, of course, different from what is used on the other networks.
In the enterprise and consumer environments, VoIP provided by service providers typically uses G.729. The network itself is typically designed for wireline access but in reality many access it over WiFi. Needless to say that network behaviour is different in each case.
Well, I guess there is no need to say too much about this subject. It is robust and voice quality is constant but at PSTN quality.
The brain and the tools that solve the problem
We have seen that different networks have different characteristics. That’s no news I guess. The conclusion though, is. Because this simple fact means that it is not efficient to decide on call setup properties and mid-call changes without knowing exactly the characteristics of the caller and callee networks.
Solving the quality issues when calling between networks requires three things:
- Knowledge of all networks’ characteristics
- Real-time monitoring information
- Voice enhancement tools that will be managed dynamically based on actual call and network information
The entity that answers these requirements would be an SBC or a transcoding GW that resides in the core of the network. It would manipulate voice using various voice enhancement tools, decide on the best codec and rate for a call between the networks and use monitoring information for its decisions
On October 8th, I’ll be speaking on this topic at the LTE Voice Summit in London. If you happen to be there, drop by and say hello.
Here is a sneak peek to one of my presentation slides.