Building the 4G Foundation

Looked at from several points of view, long-term evolution (LTE) is an apt name for the latest wireless initiative. Literally, it is the name of the 3GPP project to define the long-term evolution of the universal mobile telephone system (UMTS) to a true fourth-generation (4G) level.

Technically, the required computing power finally is available to support the use of orthogonal frequency division multiplexing (OFDM) and related variants. Protocol-wise, the packet-oriented system has reached a new level of IP complexity. And predictably, the LTE specification itself, sometimes referred to as 3.99G, continues to evolve.

OFDM has long been recognized as especially good for improving communications through less-than-ideal channels. The primary reason is the lack of interference between subcarriers that results from linking their frequency spacing to the modulation rate. The nulls of each sin X/X spectrum corresponding to square-wave phase modulation of the subcarriers are aligned to coincide with the peaks of adjacent spectra. This is a very different situation than
occurs in simple frequency division multiplexing (FDM).

LTE channels use subcarriers spaced 15 kHz apart in the uplink (UL) and either 15 kHz or 7.5 kHz in the downlink (DL). For the 15-kHz spacing, each modulating symbol lasts 66.7 µs. This obviously is much longer than a CDMA system’s 260-ns symbol length. Nevertheless, OFDM achieves a high data rate by simultaneously transmitting hundreds of symbols.

Multipath effects pose very real limits on the performance of a CDMA system. As the bandwidth and chip rate are increased, the symbol length becomes shorter and the delay variation among received data paths more important. When the variation is comparable to the symbol length, rake receivers and equalizers are needed to sort out the resulting inter-symbol interference (ISI). In contrast, the addition of a relatively simple cyclic prefix (CP) to each OFDM symbol can completely eliminate ISI up to a path difference corresponding to the added CP time.

When discussing the relationship between modulation rate and subcarrier spacing, both the time and frequency domains are involved, and this is the crux of the technical challenge with OFDM systems. The subcarriers are defined in the frequency domain, but groups of symbols are transmitted in the time domain.

In an OFDM system, the amplitudes and phases of the subcarriers correspond to the symbols to be transmitted. After the symbols are mapped to the subcarriers, the subcarriers are separately inverse Fourier transformed to the time domain, modified by the addition of the appropriate CP, and vector summed to produce the composite waveform. These operations require a lot of computing horsepower and first became commercially practical in Wi-Fi and WiMAX systems that also are based on OFDM.

The time-frequency matrix that defines an LTE transmission comprises hundreds of resource elements (RE) each one symbol long in the time domain and one subcarrier wide in the frequency domain. The smallest unit that can be scheduled for transmission is a resource block (RB), which occupies 0.5 ms and 180 kHz. Taking the CPs into account, there usually are seven symbols in each 0.5-ms slot. There are 12 subcarriers in 180 kHz so 84 REs make up an RB.

Frequency division duplex (FDD) transmissions are organized into 10-ms frames with 20 0.5-ms slots. Time division duplex (TDD) operation retains the 10-ms frame timing but split into two half frames, each with five subframes. The subframes are assigned to UL or DL operation depending on one of seven UL/DL configurations, and special functions such as guard periods and pilot time slots also are used. UL and DL use different modulation formats and synchronization reference signal structures.

Although LTE is based on OFDM, this modulation scheme isn’t used in its basic form. Instead OFDM is modified to OFDM access (OFDMA) in the DL and single-carrier FDMA (SC-FDMA) in the UL. In OFDMA, the subcarriers associated with each user are dynamically allocated according to the DL channel characteristics. This means that narrowband interference and fading can be avoided.

Separately, the UL transmission format is altered to reduce the high peak-to-average power ratio inherent in OFDM systems. Similar to OFDMA’s reallocation of subcarriers in the time domain, SC-FDMA reallocates each group of symbols within one symbol period across the 12 RB subcarriers. This results in the UL symbols being transmitted serially at 12 times the original rate but across the same 180-kHz bandwidth.

In addition to supporting FDD, TDD, and multiple frequencies and bandwidths, LTE includes various forms of MIMO operation involving two or four transmission streams and receive antennas. “Diversity techniques increase the robustness of the signal path but do not increase the data rates. Spatial multiplexing leverages the addition of transmit and receive antennas to increase the fundamental channel capacity. Suitable channel conditions are needed to make this practicable…. LTE uses multi antenna techniques dynamically, placing considerable demands on the evolved node B (eNB) and user equipment (UE) to report the correct channel state information and react to it properly.”¹

LTE Test Implications

This is not a simple system, nor is it one that’s easy to simulate. Nevertheless, the various subsystems must be tested both separately and as part of the overall system. However, just as communications networks have been evolving, so too have the chips used in their implementation.

New LTE chip sets are likely to have a DigRF high-speed serial interface linking the baseband and RF ICs rather than the traditional analog link. Clearly, an analog signal generator no longer can simulate the baseband-to-RF IC link, and you have to use a different kind of test instrument. The situation is complicated by the sophistication of the DigRF link that includes its own protocol stack.

Taking into account the mixed data and control signal traffic on the DigRF bus, “…analysis starts by acquiring the digital signal using an appropriate tool such as a logic analyzer. The baseband signal can then be analyzed with vector signal analyzer (VSA) software as if the signal had been captured as an analog or RF waveform.”²

Power measurement is another difficult area because traditional power meters and spectrum analyzers are not well suited to noise-like and bursted signals. “In most cases, the best way to measure the power of an LTE signal is to use the power calculations available in demodulation applications. The time synchronization and time selectivity of demodulation (selecting which part of the frame to measure) are important for ensuring that the desired portion of the signal is measured…. An important issue for [any] measurement approach is the use of consistent or equivalent signal content since LTE IF and RF power levels vary with signal content.…”³

Yet another problem occurs with LTE BER measurements. Baseband coding and decoding are used to minimize errors as well as enable error correction. Unless these mechanisms are correctly implemented or simulated, you can’t measure a receiver’s coded BER at the minimum RF power input. Being able to mix simulated designs and signals with real measurements and network elements is an advantage in this situation.

LTE Test Sets
One way to address these and many other complex measurement issues is to use an LTE test set. AT4 Wireless’ Chief Technical Officer Joaquin Torrecilla explained, “Modern test sets generally are based on a software radio approach, providing the flexibility to adapt to multiple standards. With a test set, engineers can perform measurements in close to real operating conditions, with all the protocol layers working and the UE communicating with a network. In contrast, many measurements done with individual instruments are restricted to scenarios where test signals are produced by the UE.”

The Model E2010 Broadband Wireless Test Set is the platform AT4 uses to run the S3110 LTE layer 1 application. The combination provides a DL signal generator, a fading emulator, and an UL signal analyzer.

Real-time protocol decoding and comprehensive triggering are among the UL signal analyzer features. Triggering is especially important because it defines the time that the specified combination of physical-layer events and external events occurred and causes the corresponding UL signal to be captured. Once the signal is captured, time, frequency, and modulation quality measurements can be made. Captured IQ vectors can be exported to MATLAB routines should further processing be required.

The company’s T4010 LTE RF Test System also is based on the E2010 platform but used as an eNB emulator (Figure 1). The tester supports functional block verification, RF parametric testing, margin search, RF performance testing, and complete system validation and certification including type approval. By itself, the T4010 is capable of performing greater than 60% of the required FDD RF test cases. Complete coverage requires the addition of a spectrum analyzer and CW generator.

Also based on the E2010, the T4110 LTE Protocol Tester supports development of LTE protocol stacks including successive regression testing. A single E2010 test set can emulate two full LTE cells typically configured as 2×2 MIMO at two different frequencies. Full test system automation is possible as well as expansion to 4×4 MIMO with interferer and additive white Gaussian noise (AWGN) sources.

Through its recent acquisition of Catapult, Ixia now provides comprehensive eNB testing. James Rankin, product manager for LTE at the company, said, “The IxCatapult system enables accurate network testing by simulating actual traffic loads at full levels in each LTE radio sector for full node B capacity. The test system generates simultaneous voice, video, and data streams for realistic network testing with incremental traffic until the eNB’s capacity has been exceeded.”

Aeroflex has developed at least two platforms suitable for LTE: the 7100 LTE Digital Radio Test Set that specifically targets LTE and the TM500 that emulates UEs based on several technologies including LTE. The most recently introduced LTE variant, the TM500 TD-LTE, supports TDD base-station development including handoff between multiple base stations.

Both the TM500 LTE model based on LTE FDD technology and the TM500 TD-LTE feature test, logging, and measurement capabilities at all protocol layers and include advanced MIMO testing. Typically during development, you need to configure parameters and scripts to execute complex test scenarios at different layers. Both local and remote test script automation are supported.

In contrast to the TM500, which emulates UE operation and is intended for base-station development, the 7100 (Figure 2) simulates the network from the physical layer to the core network IP infrastructure and includes both parametric analysis and protocol logging and diagnostics. Optionally, the 7100 can address 2×2 MIMO and fading simulation, and a second RF carrier can be used in handoff testing.

A very important aspect of testing is inter-system handover. It will be necessary for LTE networks to hand over calls originating on or intended for earlier generation networks based on GSM/GPRS/EDGE and WCDMA/HSPA technologies. This test capability is planned as a future 7100 option.

Phil Medd, product manager Aeroflex Test Solutions, highlighted some 7100 details. “The test set provides real-world LTE signal emulation capability through its built-in fading simulator feature. This allows up to nine-way multipath fading with Doppler shift simulating speeds up to 1,000 km/h. In addition, an AWGN source allows assessment of performance under challenging conditions, such as at the cell edge. By implementing this feature in baseband, accuracy and resolution are determined purely by word widths, in contrast to traditional RF-based fading solutions that require careful calibration.”

Multipath fading is a critical aspect of MIMO operation in which the different paths are used to increase the channel capacity or make communications more robust. So, it’s not surprising that separate fading instruments are available that may be more flexible than a built-in solution. For example, Spirent Communications’ Model SR5500 Wireless Channel Emulator has two RF channels, each with 24 paths.

Spirent also provides the 8100 Mobile Device Test System intended for UE development and test
(Figure 3). The SR5500 is part of this network and channel emulation tool configured from separate modules: data throughput, data call reliability, voice call reliability, video, and three A-GPS conformance modules. For performance monitoring, the 8100 modules include an SR3420 Network Emulator with network-grade protocol stacks.

According to Graham Celine, senior director of marketing at Azimuth Systems, “We believe in a hybrid solution—a test set that uses as realistic an environment as possible, including real base stations and user equipment and full bidirectional channel modeling. As a supplier of only the channel emulator component, we actively partner with many other test-equipment vendors to ensure we deliver test sets and automated test suites with the greatest benefits.”

Rohde & Schwarz also provides real-time fading simulation, in this case as an option to the company’s SMU200A VSG and AMU200A Baseband Signal Generator. Andreas Roessler, R & S technology manager North America, explained, “Besides many other standards, the fading simulator covers the LTE-specific fading profiles EVA, EPU, ETU, and the high-speed train (HST) profiles. The use of standardized conformance test fading profiles allows the test-equipment requirements and cost to be kept down.”

Keithley’s approach to LTE testing integrates the company’s VSA and VSG instruments together with SignalMeister™ software that provides a comprehensive set of signal and channel models. You create the required test files by combining icons from lists of analysis, generation, and signal-processing capabilities. Downloading the files to your actual VSA and VSG configures them for the technology you’re working with: WCDMA, Wi-Fi, or LTE.

MIMO operation is supported by combining multiple instruments and a separate synchronization module. Mark Elo, the company’s RF engineering manager, commented, “Because the number of transmitters and receivers can vary greatly in MIMO topologies, test systems need to be highly flexible to accommodate this variability. In general, a test set designed for a specific device or topology tends to be less flexible than a test system that employs separate signal generators, analyzers, and software for defining the signals and channel models.”

Agilent Technologies provides many separate instruments as well as test systems suitable for LTE, but the company also highlights its capability to mix simulated and actual system elements. For LTE, this can be accomplished by using the company’s Advanced Design System (ADS) software signal-processing models and preconfigured simulation setups. You can combine live RF simulations, baseband simulations, and real-world measurements in a physical-layer model.

“For UE development and test, the Model E6620A Wireless Communications Test Set provides a powerful, common hardware platform,” according to Jan Whitacre, Agilent’s LTE program lead and project director. “The E6620A one-box test set offers real-time, system-rate network emulation for L1/L2/L3 UL and DL via RF or digital baseband. With full RF measurement capabilities, the E6620A supports MIMO and protocol conformance testing.”

Stemming from the strategic partnership Anite and Agilent formed late in 2007, Anite recently introduced the SAT LTE Protocol Development Test Set based on Agilent’s E6620A Platform (Figure 4).

Separate Test Instruments
A number of considerations go into the choice of test equipment. A test set represents a significant investment and may not be necessary if you already have many of the constituent instruments such as VSGs and VSAs. Sometimes test sets trade speed for flexibility and really don’t offer much information beyond determining that a device meets the minimum specifications. Clearly, engineers have different test equipment needs at different points in a project.

“At the beginning [of a project],”Agilent’s Ms. Whitacre explained, “designers need to test RF and baseband parametric performance under various test conditions. They need equipment to be higher performing to fully test out each aspect of the design. They need to stimulate and analyze the different pieces of the block diagram. This is why they use individual pieces of test equipment. Separate instruments have the flexibility to move to different needs within the lab.”

Suitable Agilent equipment includes the MXA Signal Analyzer and the MXG Signal Generator. An LTE signal simulated in ADS can be downloaded to the MXG’s Arb, creating a physical test signal. The MXA in conjunction with Agilent’s 89600 VSA software captures and demodulates the DUT’s RF output, after which a number of measurements such as BER are possible via the company’s LTE Wireless Library. Which instruments are most appropriate obviously depend on the application details, and Agilent provides a wide range of VSAs, VSGs, spectrum analyzers, oscilloscopes, and logic analyzers applicable in LTE testing.

“When they start pulling the pieces together, they need to look at the overall design and want test equipment that will provide the call processing and signaling to make the device look like a single unit. You need a one-box test set that provides the simulation, analysis, and signaling to communicate with the phone,” Ms. Whitacre concluded.

R & S’ Mr. Roessler agreed, adding that the necessity to simulate the real environment requires the performance of a VSG like the SMU200A together with fading, noise, and interferers. This is provided in the company’s TS8980 Conformance Test System, which Mr. Roessler described as an addition to separate instruments rather than a distinct integrated unit.

The noise-like characteristic of digitally modulated signals is challenging for traditional test instruments. Legacy communications systems have relied on modulation rates that were similar to the information rate. In contrast, the waveforms associated with digital modulation schemes such as WCDMA and OFDM change very rapidly. This is one reason that conventional swept-spectrum analyzers are not appropriate for LTE and WiMAX testing.

Instead, an analyzer that captures a block of time-domain data in real time provides a better match to the nature of the signal. VSAs operate in this way, performing FFT-related operations on the captured data to produce frequency-domain information. This could be in the form of the signal power spectrum or presented as a complementary cumulative density function (CCDF) giving a statistical overview of the signal.

Tektronix has developed the Real-Time Spectrum Analyzer (RSA) based on VSA technology but enhanced through sophisticated DSP techniques. Once the signal has been captured, the company’s RSALTE software demodulates it in terms of an LTE resource allocation map, an XY matrix relating RB subcarriers and time slots. Maptool is a feature of the software that allows you to modify the map if you need to implement nonstandard schemes. RSALTE also provides the error vector magnitude (EVM) analysis metric and spectral measurements.

According to Li Cui, product manager at the company, “RSALTE software can analyze the frequency settling time and phase errors that occur during a burst transmission and correlate them with their impact to modulation quality. RSALTE is updated periodically, and the latest version supports the December 2008 LTE release.”

Anritsu’s MS269xA Series Signal Analyzers also are well suited to LTE signals because of the 31.25-MHz standard measurement analysis bandwidth, well in excess of LTE’s maximum 20-MHz bandwidth. Lynne Patterson, business development manager at the company, explained, “With options, the MS269xA can conduct analysis at the RB level, and analysis resolution can be set to the physical channel, RB, and symbol. EVM is available at any resolution. By pressing a single button, users can measure occupied bandwidth (OBW), adjacent channel leakage ratio (ACLR), and out-of-band spurious emissions.

“Waveform patterns can be output as RF signals from the MS269xA’s optional signal generator,” she continued. “With this capability, the analyzer supports RF receiver and transmitter evaluation. User-defined reference signals can be created and incorporated into the waveform files for transmission as well as loaded into the instrument for analysis.”

Anritsu provides the MD8483A Signaling Tester that simulates eNB operation when testing LTE UEs. It conducts end-to-end testing at DL speeds up to 150 Mb/s and UL speeds up to 50 Mb/s. In addition, the hand-held BTS Master™ MT8221B Base Station Analyzer is available with a 20-MHz bandwidth
(Figure 5). The MX370108A LTE IQproducer PC software is used to create LTE waveform patterns suitable for download to either the MS269xA or MG3700A VSG.

Summary

The good news is that there is no shortage of test equipment that can be applied to LTE projects. The bad news is that because of LTE’s complexity, you need to understand in great detail what you are expecting to accomplish when using any test tool. Of all the suitable instruments, only a few may address the particular problem you need to solve.

For example, if you’re developing a base station, you probably are more interested in simulating UEs than eNBs and vice versa. A comprehensive test set may provide both capabilities, but the integrated analysis and generation tools could have limited flexibility. This is where highly focused solutions such as Tektronix’ RSA or Spirent’s channel emulator can make a difference.

Perhaps the best advice is to hold LTE’s complexity in high regard. Several technical notes, white papers, articles, and books are available that delve into all aspects of LTE. The official www.3GPP.org website also is a good source for this kind of information as well as the latest status of the LTE standard. Finally, if, in spite of scrambling, turbo-encoding, inverse Fourier transforming, and resource mapping you still don’t feel technically challenged, there’s an advanced version of LTE (LTE-A) waiting in the wings.

Reference

1. Rumney, M. ed., LTE and the Evolution to 4G Wireless, Agilent Technologies, 2009, p. 57.
2. Rumney, p. 244.
3. Rumney, p. 245.

October 2009