Wrap up from yesterday’s VoLTE Roundtable
Author: Derek Kerton, Managing Partner of Kerton Group and Chairman of Telecom Council of Silicon Valley, @derekkerton
Thursday Oct 22, 2015
At yesterday’s Telecom Council roundtable meeting hosted by AT&T, our panelists from Amdocs, Intel, Mitel, Sandvine, and Imagination Technologies led by Dell Oro Group and interaction from participants (attendees listed here) discussed the topic of VoLTE.
The benefits of VoLTE are clear, customers will get a higher quality of voice call, while carriers can finally get beyond voice as a dedicated circuit-switched service. Network Operators will offer VoLTE as an IP-based service, meaning that any and all traffic on the RAN is IP-based. This reduces network complexity and operational costs, plus opens the road for voice services to evolve and integrate better with other IP services.
Service evolution includes the ability to seamlessly transition from voice to any other IP service, such as videoconference, send a picture or other media in a phone call, add in more callers, etc. While VoLTE is a reality in multiple networks, including SingTel, T-Mobile and Verizon, the reality isn’t quite as good as the story. The problem with VoLTE is “the sound of one hand clapping,”that is, to use VoLTE in any way, the call must be end-to-end VoLTE. That means that both phones, both RAN segments, and the core of the network all need to be upgraded in order for VoLTE to be activated, else conventional circuit-switched connections are made. This negates all the quality improvements, cost savings, and redundancy elimination. In short, VoLTE doesn’t pay dividends until it is widely distributed.
The panel discussed whether VoLTE could be useful between a cellphone and non-conventional end points, such as a web browser using WebRTC, or a VoIP client. The panel answered that the two ends of the call would use different compression and transport schemes meaning that a gateway (or Session Border Controller) would be needed in between. The gateway would ‘spoof’ an appropriate end-to-end connection both ways, and to transcode the flow. But in the normal case today, no – a VoLTE call will not be established if the other party is a VoIP or WebRTC user.
On the subject of VoLTE prioritization in a highly congested cell sector, the panelists explained that VoLTE traffic is prioritized, but not in the conventional way IP traffic is prioritized.
- Conventional Ethernets will normally tag high priority traffic, and then treat it differently in the data plane based on the priority tag. This works, so long as the different transport networks all respect the tag, and the tag doesn’t get stripped out (which in practice is just “best effort” and often falls apart in a multi-vendor environment.)
- In the case of VoLTE, the voice traffic is not tagged, but rather VoLTE traffic is given its own “bearer” in the transport plane (OSI L4), with dedicated capacity. Traffic is rated with a 3GPP QCI (QoS Class Identifier) from 13 types (more info here). In brief, instead of tagging the packet of a voice flow on a packet-by-packet basis, VoLTE tags the connection with a QCI, and sets up the appropriate bearer with a Guaranteed Bitrate (GBR). Lower priority traffic cannot access or interfere with that bearer.
If you look closely at VoLTE, you can still see the lingering notions from circuit-switching being applied to an IP world. This is a good thing: reproducing circuit-switched reliability in an IP-based architecture, and offering the best of both worlds.
Special thanks to Chris DePuy, the VP of Strategy @; Saraj Mudigonda, Senior Business Development Manager @; @, Strategy Planner at Intel; Kevin Summers, Sr. Director, Market Strategy @; Edwin Gans, Director of Product Line Management @ ; and Brent Stockman @, Product Manager at Sandvine, for sharing their insights and expertise on this topic. And, thanks to AT&T for hosting us.