Handsets are already getting to market under-performing in certain areas – and tests for certain applications are still nascent. Adding LTE to the mix will only put more pressure on the test community, as they support developers working under ever more intense pressure
The increased capability of LTE devices presents a similar increase in the degree of complexity for manufacturers bringing these products to market. Simpler designs could once be tested at the manufacturing stage; however, the costs of addressing problems late in the development cycle are likely to prove unsupportable today, and the requirement has shifted to the R&D phases to catch problems earlier.
The progress of the standardisation work for 3G LTE is based on a set of high-level requirements, the principal aim of which is to further improve service provisioning and coverage, but at a reduced cost-per-bit compared to 3G for both operators and users. All this is to be achieved within the context of an enhanced user experience, operational flexibility that covers both existing and new frequency bands, improved data rates and reduced latency. As such, 3G LTE will provide the major leap forward, all of which brings with it significant testing challenges.
The compressed timeline for LTE standards development is reflected in the schedules for LTE product development. As specifications for the LTE radio interface stabilise, equipment manufacturers have begun to work on components for LTE base stations and user equipment. Operators have to upgrade their core networks to support LTE, and they undoubtedly want to have user devices available in quantity at the time of commercial launch. Suppliers of design and test equipment have to keep pace to provide the tools necessary for these tasks. Because of the newness and complexity of LTE technology, there are a range of engineering design and test challenges to address in the different parts of the LTE infrastructure:
RF: The variable channel bandwidths specified for LTE increase the system's flexibility and capability but also add to its complexity. The use of multiple antenna configurations and OFDMA to support high data rates adds further complication, although it's expected that by the time LTE products reach the RF testing stage, test engineers will be able to apply lessons learned from implementing MIMO and OFDMA in WiMAX. However, the use of SC-FDMA in the uplink will result in some challenges unique to LTE. With performance targets for LTE set exceptionally high, engineers have to make careful design trade-offs to cover each critical part of the transmit and receive chain.
Layer 1/Baseband: To support the high data rates that are the goal of LTE, exceptionally large amounts of processing power are needed, particularly in the baseband, where all the error handling and signal processing occurs. Baseband designs will be modeled using PC simulation on both the UE and network sides, and reduced-speed emulation of hardware prototypes is also happening.
Layers 2/3: LTE Layer 2 is split into three sub-layers, including the Medium Access Control (MAC) and Packet Data Convergence Protocol (PDCP). Design challenges at this layer will be the handling of significant amounts of data in the PDCP and implementation of the 2ms MAC turnaround time. Layer 3 handles the main service connection protocols. Detailed specifications for both of these layers are still under discussion. Although early product development can be accomplished with simulation of these layers, the integrity of a device design cannot be determined until they are properly integrated with the baseband and RF sections at full operating speed.
Along with LTE-specific challenges are those associated generally with wireless design. Overall system performance depends on the performance of both the baseband and RF sections, and each is associated with particular impairments-for example, nonlinearities and noise figure in an RF up-converter or down-converter, phase and amplitude distortion from a power amplifier, channel impairments such as multi-path and fading, and impairments associated with the fixed bit-width of baseband hardware.
Not the least of all these challenges is the fact that LTE is an evolving technology, and as such is open to change and interpretation. The early availability of conformance tests will help alleviate interoperability issues and provide basic testing. However, from day one of commercial launch LTE must deliver an outstanding user experience in terms of voice quality, quality of data services, and battery life. For that reason comprehensive functional testing and real-world verification of LTE products is essential.
This in turn has put pressure on the test community as it is faced with not just testing protocols that are more complex, but also with dealing with technologies that are not stabilised in fixed standards. Nor are the challenges "merely" on the physical layer – pressure is being applied to produce tests in layer 2-3 that will ascertain the likely user experience of phones, rather than merely give a pass/ fail on.
At the moment, some European operators are reporting return rates on certain handsets of 50%, according top Spirent's Nigel Wright, Vice President of Wireless Product Marketing. But the surprising thing is, the reason for return has to with something as "simple" as handover.
"We talk to large operators in Europe, and 50% of some models are being returned – and these models are also the ones that are dropping calls the worst. And the biggest reason for dropping a call has been in handover failure," Wright says.
This is interesting because, as the industry looks forward to LTE, in all its OFDMA, MIMO, beam forming, 4×4 antenna complexity, the number one issue in returned handsets appears to be something this magazine was covering back in the early years of the decade – namely how to deal with handover between 3G and 2G networks.
Wright agrees.
"It is spectacular that this was not sorted out years ago. We were very surprised, but if you ask operators what their number one issue is right now they consistently say it's this. Most of them now keep a blacklist of handsets with performance issues."
As 3G LTE is an evolution of existing UMTS systems based on W-CDMA and will also fully integrate with existing GSM/GPRS/EDGE networks, seamless handovers will be critical to the gradual rollout of the first 3G LTE networks and deployment of the first LTE mobile devices. Such handovers might simply be inter-cell between neighboring 3G LTE cells or they could be handovers to W-CDMA or GSM/GPRS/EDGE as a user moves in or out of LTE coverage.
Wright thinks that much of the problem could be to do with a "GSM mentality", in which operators have always trusted and relied on conformance testing. In the CDMA world, where testing was less standardised, operators have long developed out their own test environments, in-house, to assure performance. So are European operators starting to take more responsibility for their own devices under management?
"They are starting to look at it." Wright says. "but I wouldn't say it's really been seen widely yet. There's still the GSM mentality of "if it's certified then it's good".
Yes even though handsets may have achieved certification, that is no guarantee of a consistent level of performance – and hence user experience – across handsets.
Here's Wright again. "It's really about achieving a minimum level of performance, of standardisation compliance. Yet in recent benchmarking for an industry analyst in the USA, we saw devices operating under identitical network conditions achieve two times the data rates of others, depending on the chipsets involved and the efficiency of the network and channel-related algorithms. That's a big difference.
"Significantly, we have started seeing large operators develop their own acceptance tests, versus accepting the terminals as they are. AT&T has put in a pretty significant back office and its own data performance testing."
But if HSPA has been showing up performance difficulties, LTE will place even more concerns on the operator and handset manufacturer side.
"Historically, new technology roll-outs have been subjected to delays, often due to problems experienced when the new technology mobile devices are tested against the new technology networks," says Aeroflex's Phil Windred.
For example, in order to obtain the all-important performance advantage for higher data rate mobile applications, the 3G LTE specifications will be based on a switch from W-CDMA to OFDM (Orthogonal Frequency Division Multiplexing) modulation technology. This represents a significant change at the very lowest level of the radio communications and achieving synchronisation will be a major challenge.
Complete visibility into the very lowest layers of the radio modem will allow users to diagnose the actual cause of a synchronisation problem rather than just knowing that synchronisation has failed.
Without the higher layer protocol, it is necessary to completely configure the physical layer using test scripts. As a consequence, many early test failures may not be the result of real problems, but rather by a mismatch in the setup between the prototype under test and the test equipment. With hundreds of parameters that need to be selected, the risk of a mismatch is significant.
A further implication of physical layer testing when the higher layer protocol is not available is that test automation is essential to ensure extensive and complete testing. The incorporation of test script configuration tools will allow the easy generation of all scripts needed to select the different configurations and tests. These various scripts can then be initiated by a test controller as required to synchronise control of the prototype under test and the test equipment. It will be possible to alter parameters in real time to enable test coverage to be extended across the wide range of different configurations used in a live system both in relation to test of the 3G LTE network and test of early 3G LTE prototype mobile devices. This will allow the early detection of software bugs associated with particular parameter values that will not otherwise be found until much later in the design cycle when diagnosing and rectifying errors tends to be much more expensive.
"With the current method of working for conformance testing, adding LTE capability means the scope of the tests required will end up being massive, and could still not cover all the performance envelope of the device," Wright says.
So how will the industry deal with the great scope of LTE device testing?
"I'm not sure how this is going to develop. Device manufacturers may be stuck with huge manadatory testing requirements, and operators cannot be sure even then if they will catch it all. It's going to be interesting to see how test approval evolves over the year. Something needs to change."