Saturday, November 19, 2011

LTE and the Future of Mobile Networking

Upperside Conferences invited me to present about the advances in voice and video technology at the Voice over LTE conference last week. Upperside's conferences always focus on a specific technology subject and perfect if you want to learn "everything" about it. This time the talks were all about LTE, which stands for Long Term Evolution (LTE), and is a key new technology that revolutionizes mobile networks.

LTE versus VoLTE

LTE is a radio access technology that was developed in 3GPP to enhance performance and efficiency in mobile networks, more specifically, increase bit rates (up to 150 Mbps), improve cell spectrum efficiency, and reduce air interface latency. Since LTE itself is a pure IP packet service, transporting voice, SMS, and other media like video and IM has to be specified separately.

Voice over LTE is defined as an end-to-end service that includes not only the efficient LTE radio but also the IP Multimedia Subsystem (IMS) core, the Evolved Packet Core (EPC), and LTE capable devices (dongles, tablets, and smart phones). The key difference between LTE and VoLTE is therefore that LTE is the wireless interface while VoLTE is a complete solution for transporting voice and SMS over the new network.

The problem is that the solution – although fairly new – already needs extension with IM, video, and other functions, and the "voice" in VoLTE is very limiting. So the mobile networking community created something called Real-time Communication Services extension (RCS-e), that extends the voice and SMS services with, for example, IM/chat, file transfer, image and video share. It is roughly the equivalent of what we call Unified Communications in the enterprise communication space. Similar to UC, RCS-e is based on service/capability discovery and uses the SIP protocol. The role of SIP in enterprise UC is described here.

As with any other technology, getting the acronyms right is half of the work, so here are the important ones.

IMS is an architectural framework for delivering multimedia services. The idea is decouple the application from the access method, so that the same functionality can be accesses over wireless (3G, 4G) and fixed networks. This idea is not new: web applications, for example, do not care about the underlying network as long as it carries HTTP; this is probably why they are dubbed Over The Top (OTT) applications in mobile networking lingo. IMS is a revolution in mobile networks where traditionally separate functions were implemented for each type of network. By developing applications only once in the IMS core mobile SPs can shortens time-to-market and compete more successfully with OTT applications. IMS is also designed to provide QOS for applications through the different access networks, and this is a major differentiator for mobile SPs.

EPC is basically an IP router with mobility intelligence that includes handover (switching the connection when the mobile device moves from one radio cell to another), roaming (provides access when users leave their own SP network and enter the network of another SP), etc.

From Circuit Switched Voice to VoLTE

The LTE radio technology is different from 3G and building LTE networks requires substantial capital investment, as pioneers Verizon and MetroPCS in the USA and Vodafone in Europe know very well. As a result, LTE networks will coexist with 2G and 3G networks for long time, and it is critical to find a way to switch calls seamlessly from LTE to non-LTE networks when the user leaves LTE coverage. There are two ways to do that - Single Radio Voice Call Continuity (SRVCC) and Circuit Switched Fall Back (CSFB) – the former using single radio and being less expensive, the latter using dual-radio and costing more.


In terms of bandwidth, LTE can theoretically provide up to 150Mbps shared in a radio cell. LTE bandwidth is symmetric, that is, upstream bitrate can be equal to downstream bitrate; this makes LTE best choice for symmetric services such as online gaming and real-time voice and video. If there are 200 users in the radio cell, each user can get 0.75Mbps which is enough for high-quality H.264 video.

A bigger QOS problem in LTE networks is network congestion that leads to rapidly increasing delay (and packet loss) with very little advance notice. In video conferencing, the receiving endpoint sends congestion notification to the sender (SIP does that via RTCP while H.323 uses the H.245 Flow Control message), and the sender down-speeds, that is, reduce either resolution or frame rate. Mobile networking vendors are researching ways to detect congestion on a radio cell level; that would require the radio node to send congestion notifications. Since the radio node sees all traffic from all users in the radio cell, it can give an advance warning. It will be important for mobile UC application with video capabilities to listen for such notifications and down-speed – even if their own session is performing well.

Packet loss in LTE networks usually becomes a problem when the user is at the periphery of the cell where the radio signal is weak. When the signal strength is good and the user does not move, frame error rate can be as low as 0.2% which results in negligible packet loss.

To provide Quality of Service to applications, VoLTE defines QOS Class Identifiers (QCI); each of them is appropriate for certain type of traffic. For example, QCI1 bearer is optimized for VoIP/VoLTE. Tests show transmission latency of 140-160ms, which has to be added to the RTP delay and voice/video codec delay. The resulting end-to-end latency can therefore be higher than 200ms, and more work has to be done to reduce the latency below the 200ms limit critical for interactivity on voice and video calls.

Another important QCI for real time communication is QCI5 that is used for signaling (call setup/tear down). Currently, the end-to-end call setup time in 2G/3G networks is about 6 sec, and VoLTE performs better: call setup times measured over the QCI5 bearer are about 2-3 seconds, even when the LTE device is in battery saving mode.

The QOS issues in mobile networks listed above are not very different from the QOS issues in fixed IP networks a decade ago. Video conferencing technology has matured over longer period of time and has therefore already implemented mechanisms for compensating bandwidth reduction (for example, down-speeding implemented in Polycom HDX video endpoints), packet loss (for example, Polycom Lost Packet Recovery), lip sync, etc.

LTE as a Fixed Network Replacement

I often hear that LTE is not a substitute for fixed access networks such as VDSL and FTTH, and my reaction is always "Why not?" As I learned at the VoLTE conference, there is a business case for using LTE instead of DSL to provide high-speed network access. For example, DSL providers in Germany pay a fee of 10 Euros for using the last mile of copper, and Internet SP are very eager to get rid of this cost. Since LTE can provide bandwidths comparable to fixed lines, modem vendors are adding LTE to the portfolio of access technologies. A great example is the FRITZ!Box, the most popular home gateway in Germany, that combines a modem (DSL, cable, and since October - LTE), a router, a firewall, a Wi-Fi access point and a DECT base station. It is a perfect solution for people like me who hate cables lying in the living room or office. Reported throughput over LTE is up to 100Mbps downstream and 50Mbps upstream which makes it Category 3 LTE device.

I have already heard that some service providers in the USA are experimenting with LTE as a fixed line replacement for services to businesses. Considering the cost and time necessary to setup a fixed line (like T1 or T3) to the customer premises, using LTE is a very attractive alternative for service providers. Now imagine that the business has an IP-PBX that uses SIP trunking to connect to a service provider. Bundling LTE and SIP trunking services is suddenly not a far-fetched idea.

All in all, I think LTE will impact both residential high-speed access and business services provided by service providers. The only "if" is Quality of Service. According to the discussion at VoLTE QOS mechanisms in LTE perform well in test environments and initial field deployments but will they do a good job once the LTE network is flooded with LTE capable devices that compete for resources?


The great thing about LTE is that it makes the mobile networks like any other IP network. Enterprise IP applications that used to require complex gateways to interface to mobile networks will now be able to run on the mobile network without much customization. The low hanging fruit is running Unified Communication soft clients like Polycom Real Presence Mobile on media tablets and leveraging the LTE network to connect back to the enterprise network. Going IP end-to-end will help reduce complexity but also cut latency to the absolute minimum. More work will be required in the area of QOS, especially in the congestion detection and notification on a radio cell level.

Mobile Service Providers will continue to develop IMS based application, also leveraging RCS-e, and it will be interesting to track the adoption of IMS applications and other, so called OTT, applications. While most of the Video Networker followers are from enterprise background, and are therefore familiar with the efforts in the enterprise UC environment, we should not ignore the efforts in the mobile networking space to solve the fundamental UC problem.