By Fabio Tignanelli, consultant in communication and electronic systems, PA Consulting Group
As operators embark on the roll-out of trial LTE Networks, so the 3GPP standards body submits first specifications for LTE-Advanced (LTE-A) - the first truly 4G standard and the next attempt to satisfy the ever increasing thirst for bandwidth. How can infrastructure, handset, and test equipment vendors ensure a timely and cost-effective transition from LTE to LTE-A? Fabio Tignanelli, of PA's Communications and Electronics Systems team, offers some insight on the challenges behind implementing LTE-A.
LTE Advanced (LTE-A) is the 3GPP answer to provide a true wireless broadband system in a wide variety of environments. The standard is currently being developed as part of the Release 10 of the 3GPP specification and it represents a natural evolution of 3GPP Releases 8 and 9 (LTE).
LTE-A will meet, and in some instances exceed, the requirements for IMT-Advance (IMT-A) specified by ITU as official "4G" technology. A first set of 3GPP requirements on LTE-A was approved in June 2008 and a complete study was then submitted to ITU-R as candidate radio interface technology (RIT) for IMT-A in September 2009.
LTE-A encompasses a collection of techniques designed to increase the bandwidth available to a mobile user. As headline figures, LTE-A boasts peak data rates of 1Gbps and a capacity three times higher than that of LTE. The main techniques that LTE-A employs are:
Whilst targeting a significant higher data rate than LTE, LTE-A is being standardized with the aim of maximizing IP re-use of LTE. This should reduce risk, shorten development times, and in theory reduce the capital cost of migration from LTE to LTE-A.
But this migration will not be straightforward. Specific processing bottlenecks in LTE designs may limit the re-use of this IP, which could force costly platform changes.
The key: scalability
But for infrastructure-, user-, and test-equipment manufacturers alike, the key to a successful transition from LTE to LTE-A is simple. The answer is scalability. Vendors must plan for scalability in LTE developments today, to ensure an efficient progression to LTE-A tomorrow.
We identify and discuss three key areas where the principles of scalability must be applied to ensure a successful transition to LTE-A.
LTE-A will re-use independent and scalable LTE L1 designs to serve multiple bands and enhance peak data rate.
Parallel and independent LTE L1 coordinated at higher layers
Bandwidth Aggregation in LTE-A specifies the transmission and reception of data across multiple bands. The specification allows for a maximum bandwidth allocation of as much as 100MHz. Since we cannot expect contiguous spectrum allocation up to 100 MHz, LTE-A is being designed to aggregate a mixture of both contiguous and non-contiguous component carriers (CC).
This requirement has a significant impact on how LTE platforms should be designed today so that they can be reused in LTE-A.
For example, the approach of using multiple independent PHY layers using multiple, discrete radio channels with aggregation and advanced radio control at the higher layers is likely to form the basis of early LTE-A platforms. A later cost reduction phase will focus on the integration of these elements into a single chipset. This approach is particularly applicable to eNBs and CPEs. Here, more relaxed space and power constraints allow an LTE platform to be upgraded to LTE-A with the addition, post-deployment, of extra L1 modules. This approach assumes, of course, that the original LTE platform is designed to scale horizontally.
With today's demands for tighter integration, we often see the migration of much of the radio interface into the same device as the digital baseband. While excellent for controlling cost and power consumption in LTE, this approach may mean that the new filtering and radio control required for LTE-A channel aggregation is a less straightforward addition. Here, therefore, is a good example of where including general purpose interfaces in your LTE baseband device design might greatly accelerate the development of your initial LTE-A platform. It's important not to get caught out by too much integration in LTE. With a migration from LTE to LTE-A in mind, designers should resist some of the pressures to perform to tighter and tighter systems integration, but rather design their LTE platform with a clear partition between Digital BB and RF control. This will enable the RF to be upgraded without having to substantially modify the digital baseband logic.
Building a Scalable Layer 1
Advanced MIMO techniques such as beam forming combined with spatial multiplexing within different beams will contribute to significantly increase the user data rate. In fact, LTE-A allows up to 8x8 MIMO in DL and 4x4 MIMO in UL. Furthermore, LTE-A also introduces the use of Multiple-User MIMO in the downlink.
The use of these new MIMO techniques will have a significant impact on LTE-A baseband design, where there will be a requirement to move, and process, substantially more data than before.
In PA's recent DSP-centric LTE implementation, we identified key bottlenecks (e.g. MIMO INVERSION) in the digital base band and it is clear that these choke points will only get worse in LTE-A.
The relatively recent introduction of more advanced platforms (i.e. multi-core devices) will scale to the levels of overall performance required. But the challenge of making full use of this processing power should not be underestimated. Being multi-core ready will require a design that permits key functions to be parallelized across all the available cores. Adopting this approach today will allow LTE vendors to ensure an easier transition to LTE-A tomorrow.
With regards to other kind of platform architectures, PA has recently worked on less DSP-centric development for LTE where the most processing power onerous functionalities were moved more and more into dedicated hardware (e.g. FPGA-centric platform) in order to match the required performances.
Today's LTE designs must anticipate the new software interfaces required by LTE-A.
A modular design for your LTE software today will ensure an easier enhancement to LTE-A tomorrow where innovative solutions such as CoMP, Relay and asymmetric UL and DL bands will be introduced
CoMP and Relay Interfaces
LTE-A will introduce new solutions such as Coordinated Multipoint (CoMP,) and Relay. CoMP technique co-ordinates different eNBs in order to increase the throughput and signal strength to the same UE. Further, this serves to minimize the inter-node interference to different UEs. Relay techniques, in essence, make use of intermediate nodes to relay frames so to extend the reach of a specific eNB.
It is evident that the communication between nodes (eNBs or Relay) in LTE-A will have to provide a very tight synchronization and use new interfaces. Such interfaces demand a decoupling of the external software interfaces and the internal data processing within eNBs, allowing future addition of a variable number of inter node (eNB or Relay) channels for LTE-A.
Once again, then, a well designed LTE implementation will find migration to LTE-A much easier here.
Protocol Stack partitioning
For an easier transition to LTE-A, the LTE protocol stack should incorporate clearly defined vertical (layer-to-layer) and horizontal (Up and Down Link) boundaries.
For example, LTE-A extends some ARQ/HARQ mechanisms from MAX and PHY to RLC. Thinking how to modularize the HARQ operations within MAC and PHY in LTE (vertical partitioning) will help to minimize software architecture changes for LTE-A.
Similarly, using separate definitions for PHY resource block numbers in UL and DL (currently of the same size in LTE) would decouple UL and DL in the stack. This is an example of horizontal partitioning. Since LTE-A supports asymmetric bands for UL and DL, taking this approach in an LTE design will make migration to LTE-A easier.
Enhance monitoring capabilities in eNBs to meet LTE-A test coverage
It comes as no surprise that the increased complexity of LTE-A introduces some big challenges in the areas of test and validation. LTE-A offers almost complete flexibility in mapping channels and data to different areas of the spectrum and then selecting from a library of SISO and MIMO modes in order to transmit them. The range of choices on this LTE-A protocol menu is quite extensive!
Equipment vendors, then, face a very significant increase in the number and complexity of test cases required to prove an LTE-A system.
This complexity means that achieving 100% test coverage in a lab will be much harder - probably impossible. Where it will be effort costly to reproduce complex scenarios in a lab environment, the solution will undoubtedly be the use of field trials filling in the test coverage gap. Developers should be adding fault and performance logging in LTE eNBs now in order to fully benefit from field trials for LTE-A.
Testing something as inherently distributed as CoMP presents a number of interesting challenges to all parties in the test arena. A carefully design test scenario will require mimicking of partner basestations and possibly even modeling of the CoMP interface of different vendors' eNBs for interoperability tests to be valid.
On the UE side, testers must mimic at least two eNBs (with different radio environments for each) and model the control and data plane interactions to both, while extracting data from the device.
LTE test equipment vendors must anticipate these complexities in today's designs to ensure they meet the market window for LTE-A test equipment. It's never too soon.
LTE-A is the 4G mobile standard proposed by 3GPP. The demand for LTE-A is real, and the standards bodies are working hard to realize the specification.
But success of LTE-A will demand a costly and timely effective migration from LTE. A smooth migration to LTE-A tomorrow will only be possible with scalability built-in to LTE designs today. This scalable approach permeates all aspects of system design - from platform and architecture through to system test.
To visit PA's communication and electronic systems pages, please click here.