Tackling MIMO Design and Test Challenges for 802.11ac WLAN
The proliferation of wireless LAN into new applications, outside of the traditional email and browsing applications, is driving the need for higher data rate throughput. New applications, such as wireless display, HDTV streaming/distribution and rapid upload/download of data, are driving two new IEEE WLAN standards requirements for very high throughput: 802.11ac for frequencies below 6 GHz and 802.11ad for the 60 GHz band.
A goal of 802.11ac is to support wireless distribution of multiple multimedia/data streams with data rates of at least 1 Gbps in the 5 GHz band. Some key features for increasing data throughput are wider channel bandwidths (that is contiguous 80 and 160 MHz, or non-contiguous 80 +80 MHz), higher-order modulation with optional support for 256 QAM and multiple-input multiple-output (MIMO) support for multiple spatial streams using multiple antenna techniques. Pre-distortion and beamforming are some further possibilities for high-performance systems.
This challenging combination of wider bandwidths, higher-order modulation and MIMO introduces new design and test challenges for the system engineer. Supporting higher-order modulation formats such as 256 QAM will require better (lower) error vector magnitude (EVM) performance from the transmitter to meet the overall system performance. Achieving a lower EVM can require better linearity from the transmitter’s power amplifier (PA), lower phase noise (in dBc/Hz) from the local oscillators (LO) being used for frequency upconversion in the transmitter, and reducing IQ skew and gain imbalance from the IQ modulator. With these tighter design margins, the system engineer may need to gain insight into predicted performance early in the design cycle to perform system design budget trade-offs. Once in the R&D and prototype testing phase, it is useful to measure the EVM performance at various stages along the transmitter chain to measure error contributions and optimize the overall system EVM performance.
This article shows how the system engineer can use system simulation to help understand the design performance and design requirements needed to achieve transmitter system-level metrics such as EVM. System performance budget trade-offs can quickly and easily be performed with system design simulation tools before designing hardware. Once hardware is available and the prototype R&D testing phase has begun, system simulation tools can then be combined with arbitrary waveform generators (AWG) to generate wide-bandwidth 160 MHz waveforms for MIMO testing. Multi-channel, phase-coherent, high-performance digital oscilloscopes with vector signal analysis (VSA) software are then used to perform the SISO or MIMO demodulation to evaluate the transmitter’s performance. In addition, the error contributions along the transmitter chain (IQ, IF, RF) can be measured using a digital oscilloscope with VSA software to help the system engineer optimize the design performance of the prototype hardware.
802.11ac MIMO System Simulation
Figure 1shows an 802.11ac MIMO RF transmitter design modeled in simulation. The spatially multiplexed 802.11ac MIMO source is modeled on the left with complex baseband outputs, a two-channel MIMO RF transmitter with IQ inputs is modeled in the center and a VSA simulation measurement is used at the output of the RF transmitter to measure the simulated MIMO performance of the modeled transmitter. Various transmitter design impairments are being modeled such as IQ modulator phase imbalance, LO phase noise (dBc/Hz vs. frequency offset), impairments from IF/RF filters and PA nonlinearities (1 dB compression point).
The simulation results for a 160 MHz, 256 QAM MIMO simulation are shown in Figure 2. The two constellations are shown on the upper left, the two spectrums centered at 5.8 GHz are shown on the lower left and the EVM is shown on the right. EVM is approximately -33.3 dB for channel 1 and -35.8 dB for channel 2, as a result of the design impairments being modeled. The modeled design impairments are impacting the channel 1 (Ch1) EVM more significantly than the channel 2 (Ch2) EVM for this example.
The system engineer can easily change parameters, such as bandwidth, to evaluate a 20, 40, 80, or 80 +80 MHz configuration. The modulation order can be changed to evaluate different QAM formats. The various design impairments being modeled in the simulation design (IQ modulator phase imbalance, LO phase noise, IF/RF filter impairments and PA non-linearities) can easily be evaluated and modified by the system engineer to better understand design requirements and trade-offs to meet the overall system-level EVM metric.
Figure 3 shows the same design as in Figure 1, with parameters configured for 160 MHz, 64 QAM MIMO. Several of the design parameters, such as phase noise and PA 1 dB gain compression, have been modified to meet the EVM design performance metrics. Several VSA simulation measurement elements have been placed along the transmitter design at the IQ modulator outputs, the IF/RF mixer outputs and the PA outputs, to gain insight into the incremental waveform impairments along the transmitter chain.
The constellations at the IQ modulator outputs are relatively clean and the EVMs are approximately -35.8 dB for Ch1 and -43.3 dB for Ch2 due to the design impairments being modeled. Only one of the two constellations is shown due to space constraints. The Ch1 IQ modulator is introducing more waveform distortion than the Ch2 IQ modulator. The mixer output used to upconvert from IF to RF shows additional waveform distortion from the LO phase noise and other mixer impairments. The EVMs are approximately -31.8 dB for Ch1 and -33.9 dB for Ch2. The PA output shows even more waveform distortion from the PA gain compression. The overall transmitter output EVMs are approximately -28.6 dB for Ch1 and -28 dB for Ch2.
Combining Simulation and Test to Create and Analyze MIMO Waveforms
Simulation can easily be leveraged in the R&D testing phase with test equipment links, which are seamlessly integrated into the simulation tools. The 802.11ac MIMO simulation source used to design the RF transmitter in the previous examples is now used to download MIMO waveforms to two arbitrary waveform generators. The complex IQ simulation waveforms are fed into two signal downloader sinks, which download the simulated IQ waveforms to two AWGs (see Figure 4). In addition, the signal downloader sinks set and configure parameters on the two AWGs, such as the master/slave relationship using parameters on the sink and a math script.
A picture of the 802.11ac MIMO test equipment setup is shown in Figure 5. The two master/slave AWGs are shown on the upper-left, two MXG signal generators for converting the baseband signals to RF are shown in the lower-left and a high performance 13 GHz digital oscilloscope with VSA software is shown on the right. A special sync cable is used on the rear panels of the two AWGs to set the master/slave configuration. The two AWGs output analog IQ signals are connected to the external IQ inputs of the RF signal generators to modulate the IQ waveforms on the 5.8 GHz carriers. The two modulated 5.8 GHz carriers are input into two of the four phase coherent channels (Ch1 and Ch3) on the 13 GHz digital oscilloscope to perform the RF MIMO measurement with the VSA software. Proper synchronization of the two MIMO waveforms on the master/slave AWGs is verified using a 13 GHz digital oscilloscope, as shown in Figure 6.
A close-up of the RF MIMO demodulation results at 5.8 GHz is shown in Figure 7, playing back a VSA recording captured on the digital oscilloscope. The two constellations are shown on the upper left, the two spectrums centered at 5.8 GHz are shown on the lower left and the EVM is shown on the right. The EVM is approximately -40.4 dB for Ch1 and -40.6 dB for Ch2 with the equalizer training set to “preamble, pilots and data.”
With the equalizer training set to “preamble only,” the measured EVMs were approximately -36.1 dB for Ch1 and -36.4 dB for Ch2. When the equalizer training is set to “preamble, pilots and data,” the equalizer estimate is averaged over all the symbols in the measurement, producing a more accurate and less noisy equalizer estimate. For 802.11ac, the typical change in EVM is between 2 and 4 dB.
The multi-channel, phase-coherent capability of the high-performance digital oscilloscope also enables the system engineer to perform measurements at various stages along a transmitter chain (IQ, IF, RF) to help debug issues that may arise in prototype hardware.
Achieving high data rate throughput for the next generation WLAN applications introduces new design and test challenges for the system engineer in terms of wider channel bandwidths, higher-order modulation formats, such as 256 QAM and multi-antenna MIMO support. This article showed how system-level simulation can help to gain insight into performance requirements and design trade-offs needed to address these new design challenges. Once the R&D testing phase has begun, simulation can be combined with test equipment using integrated test equipment links for hardware prototype testing.
A MIMO test example was shown, using simulation to generate the waveforms, which were then downloaded to two master/slave AWGs to turn the simulated IQ waveforms into physical IQ waveforms for hardware prototype testing. Two signal generators were used with the AWGs to create the modulated 5.8 GHz test signals, which were then measured using a high-performance 13 GHz digital oscilloscope and VSA software. MIMO EVM measurements were shown and the impact of equalizer training on EVM was discussed.
The combination of simulation, AWGs, and signal generators provides a flexible waveform and signal generation capability to help the system engineer tackle the next generation WLAN testing challenges. In addition to the basic MIMO signal generation capability shown in this article, waveforms with modeled impairments could have easily been created and downloaded to the AWGs to test “what-if” scenarios and stress-test prototype hardware. Using a high-performance multi-channel phase-coherent digital oscilloscope enables MIMO signal analysis, as well as the ability to measure performance at various stages along a transmitter chain.
“Testing New-Generation Wireless LAN,” Agilent Technologies Application Note,
Greg Jue is an applications development engineer/scientist in Agilent’s High Performance Scopes team. Previously, he was with Agilent EEsof, specializing in SDR, LTE and WiMAX™ applications. Jue wrote the design simulation section in Agilent’s new LTE book and has authored numerous articles, presentations and application notes, including Agilent’s LTE algorithm reference whitepaper and Agilent’s new Cognitive Radio whitepaper. He pioneered combining design and test solutions at Agilent Technologies, and authored the popular application notes 1394 and 1471 on combining simulation and test. Before joining Agilent in 1995, he worked on system design for the Deep Space Network at the Jet Propulsion Laboratory, Caltech University.