TD-LTE: A Driving Force for Mobile Broadband

At the LTE Forum at MWC 2013, Bill Huang, general manager of China Mobile Research Institute, shared with us China Mobile’s views in TD-LTE deployment. China Mobile has been actively promoting the use of TD-LTE technology.

Today, I will share with you China Mobile’s experience in deploying, trialing, and testing a TD-LTE network. We have been doing testing for the past two years, and we are pushing for large-scale rollout this year. So I want to take this opportunity to share with you some of the stuff we’ve found out.

TD-LTE Deployment Update

We are all aware that LTE now is a phenomenon. It is now probably the most rapidly developing technology with the shortest introduction cycle in the history of mobile communication. Why? Because it provides the lowest possible technology to get bandwidth to users. Today, we have more than 150 LTE networks, 35 of which are TD-LTE networks. We started trialing TD-LTE in China in 2009 and put it into service in more than 7 cities in 2011. By the end of 2012, we had 20,000 base stations covering 13 cities.

It’s fair to say today that TD-LTE and FDD-LTE network infrastructure technology are basically equal. Not only that, there are perhaps more vendors providing TD equipment as a result of legacy WiMAX. Many of you may be aware that TD-LTE and FDD LTE is highly integrated today. The only difference between them is the radio front end. The baseband and core network is the same, and in some cases, antennas are shared. 

Performance Designed for Mobile Broadband

Let’s talk a little about performance. Using 3GPP methodology, NGMN guidelines, and GCF test methodologies that we have developed over the years, we are testing the network in application situations that closely approximate those in real life. In the required specifications, I want to address the issue of asymmetric uplink and downlink configuration for TD-LTE.

Internet key performance comparison: Throughput

By configuring three downlink timeslots and one uplink timeslot, we are able to validate TD-LTE networks. In terms of spectrum utilization, TD-LTE is about 1.5 times better than FDD. In other words, you could generate 1.5 times more revenue using the same spectrum allocated for LTE. That is very important for operators that are data-centric because LTE is not necessarily made for providing voice communications; it is made to provide internet connection. If you allocate 20 MHz to deliver LTE service, with FDD-LTE, you have to divide this by ten and add ten. So, the peak speed will only be 75 Mbps. But in reality, this may be somewhere around 20 Mbps. In our testing, we found that by asymmetrically allocating bandwidth, you can achieve much higher speed because, in this case, the carrier bandwidth is 20 MHz for TDD.  Theoretically, you could achieve more than 110 Mbps, but in testing, we achieved 41 Mbps on average. TDD technology is very promising when it comes to LTE.

Internet key performance comparison: Latency

Second, there are some theoretical comparisons between FDD and TDD latency, which is somewhere around 2~7 ms because of the frame switch time in the air.  In actual testing, we determined that the average delay between the two radios is pretty much the same. Because of the various optimization points used in the system, the 2~7 ms theoretical delay is eliminated. Most importantly, we achieved less than 30 ms delay for both TDD and FDD networks, which is critical because most of the internet applications for interaction (e.g. gaming) require less than 100 ms delay. The guideline for LTE design specifies less than 50 ms delay. I can confidently say that LTE is comparable in performance to fixed broadband.

Network Requirements for TD-LTE

In 3G networks, two areas that are reasonably (but not thoroughly) addressed are video and gaming. In addition, there’s also video conferencing, which is very difficult on a 3G network because of latency.

LTE supports diverse mobile internet services

We have not only tried real conferencing on LTE networks, we have also tried live high-definition video broadcasting. We put a broadcast camera in a stadium or along a marathon route and used LTE as the only uplink mechanism during the live broadcast. In this scenario, LTE performed extremely well. In fact, we optimized the network so that we could provide real coverage along the side of the sports stadium or along the marathon route. It is very likely that you will be able to achieve broadcast-quality coverage for a sports event. This further proves that LTE is ready for prime time.

We are not just using an LTE network to provide internet access. We are also exploring the use of LTE to deploy private IP services. To deploy future carrier-class public IP services, we believe in the E2E QoS guarantee provided by LTE. We should be able to provide leased-line-quality services to future users over LTE infrastructure. This is very important for generating additional revenue for the LTE network.

Optical backhaul requirement

LTE deployment requires a reasonable overhaul of the transmission network. For those of you who have been contemplating network migration, packet transport network (PTN) technology is very important. We are now doing a massive overhaul of our transmission network, migrating from SDH to PTN. We are also massively deploying passive optical network (PON), especially in the high bit rate (10 Gbit area) in order to backhaul our signals for C-RAN. This will be the foundation for deployment of a good LTE network. It’s very important to make our operator friends here aware that investment in LTE involves investment not only in radio equipment but also in the optical backhaul.

More and more capex of ours is now invested in optical backhaul for the LTE network. There is a dramatic difference in bandwidth between the present backhaul and traditional backhaul. Backhaul bandwidth for GSM is very small, and for 3G is reasonably big (but not very big). However, larger backhaul bandwidth for LTE, especially when moving to C-RAN, is something that we need to be prepared for. Of course, the long-term saving of capex and opex in an LTE network with C-RAN technology is significant. So investment in this area is definitely called for.

Nanocell: Ideal technology for integrating LTE and Wi-Fi

Nanocell is another new technology we are moving into. We have conducted some studies of traffic patterns and have noticed that 1% of the area covered by the network perhaps accounts for more than 50% of traffic flow. This is the reason Wi-Fi has become such an important alternative mechanism for traffic offloading. In the LTE area, this technology is not only providing LTE coverage for indoor and hot zones but is also providing integration with Wi-Fi. This is so-called nanocell technology. It’s not a picocell or a femtocell but somewhere in between which carrier is going to deploy. So we call it nanocell. It further flattens the network architecture and allows an operator to collaborate with enterprises or fixed-line broadband service providers by integrating LTE cell and Wi-Fi AP into the same backhaul. Both local breakout and centralized EPC routing work together at the same time. This nacocell technology will become significant over the next three to five years.

How to build an E2E IPv6 LTE network

I know many of you are contemplating IPv6, but due to the lack of compatibility between IPv4 applications and IPv6 applications, either you settle for a dual-stack network running both IPv4 and IPv6 or you settle for IPv4 networks. Very few operators have really tried to move forward with an E2E IPv6 network. However, since LTE services have been introduced to the field, it becomes more important to have more and more IP addresses to reach end devices. So how do we do that? New technologies are now available to solve these problems. These technologies may be BIH or newly emerged dual translation. What will these technologies do? They will embed the ability to provide backhaul compatibility for IPv4 applications. In other words, IPv4 IP stack will be compatible in an E2E IPv6 network. These technologies are becoming very important, and without them, it’s very difficult to deploy an E2E IPv6 network.

Carrier IP public service (CIPS)

Let me talk a little bit about carrier-class IP public services. What is carrier-class IP public service? To start with, we know that the internet is a public IP service, but there is no QoS guarantee, and you cannot use it to provide SLA to any subscriber. Today, we deploy leased line; we deploy virtual private networks, but it’s for enterprises that use VPNs or leased lines communicate with each other. This is where carrier-class public IP services come in. In this new network, we provide customers with public IP services but not internet access. It’s a public IP services with QoS and SLA. We need to have operators to work together to provide interconnection of such a network. Inside our own carrier network, we can provide the service, but to provide the service across the carrier boundary, we have to build a new IP network that will serve as a public IP network extension. This is the best guarantee to continue to maintain the internet transparency. 

Terminal Recommendations

Last, I will talk a little bit about terminals. If you look at the technology today, we are at the verge of breaking out, just like in the 1980s, when all the different types of PCs finally became IBM PC compatibles. Today, we have many different spectrums, technologies, and countries with different radio requirements. Mobile phones, although they look exactly the same, are actually customized for every different country. Now we are in the era where new technology is introduced in both the semiconductor and unified protocol stack. We are entering the era where mobile phones are becoming global.

At the TD-LTE GTI Summit, we will introduce eight new types of devices that have global roaming. We will introduce five modes and ten different bands. We hope to promote this technology to make sure that operators, big or small, enjoy a unified device market and low-cost mass market LTE terminals. This is the future. I hope that all operators issue similar requirements, and we have white papers available for our device manufactures. Maybe in the next two to three years, these types of devices will not just be for the high-end market but also for the prevailing mid-end and perhaps even the entry-level segments.