While LTE and LTE-Advanced are still being deployed, research into next generation wireless networks is already in high gear. That next generation – 5G – will likely comprise a dense, highly integrated network of small cells supporting peak data rates of up to 10 Gbps and 1 ms or less roundtrip latencies, while utilizing a number of different air interfaces at both microwave and millimeter wave frequencies. The combined network may be able to support everything from simple machine-to-machine (M2M) devices to immersive virtual reality streaming. While that bodes well for an end user experience, it presents some interesting challenges for those engineers developing 5G systems.
Figure 1 Flexible 5G waveform generation and analysis testbed.
Figure 2 Signal creation software for creating custom, wide bandwidth OFDM signals.
Making the leap from prediction to practical implementation starts with the creation, generation and analysis of prototype signals. As there is no 5G standard at this time, no physical layer waveforms have yet been defined. Although there is a lack of consensus on 5G waveforms, filter bank multi-carrier (FBMC), universal filtered multi-carrier (UFMC) and orthogonal frequency-division multiplexing (OFDM) waveforms are being considered. Other potential candidates include waveforms at sub-6 GHz frequencies and those at microwave and millimeter wave (mmWave), which may involve wide bandwidths of up to 2 GHz. The number and variety of waveforms, frequencies and bandwidths being researched introduce new test challenges for 5G signal generation and analysis.
Key to overcoming these challenges is flexibility during 5G research and early testing. Engineers must have the ability to perform “what-if” analyses while they are evaluating early concepts and new candidate 5G waveforms. Without this ability, the risk of choosing the wrong path and not discovering issues until much later in the development cycle – when it is much more costly and time consuming to change – can increase. Flexibility, especially with signal creation and signal analysis tools, can be especially important, as they enable rapid changes in direction as strong 5G waveform candidates emerge. Engineers also need the flexibility to use a wide range of modulation bandwidths (from several megahertz to a few gigahertz) and frequency bands (from RF to microwave to mmWave).
To address these challenges, engineers would ideally like to combine off-the-shelf hardware and software to create a flexible 5G waveform generation and analysis test platform, such as Keysight’s 5G waveform generation and analysis testbed reference solution (see Figure 1). The testbed reference solution provides flexibility through its software and hardware elements. In the software, flexibility ensures that engineers can generate and analyze various types of 5G candidate and custom waveforms. With the hardware, both flexibility and scalability work together to give engineers the ability to generate and analyze signals from RF to mmWave frequencies with up to 2 GHz bandwidth.
Figure 3 Custom OFDM signal with ~1 GHz bandwidth at 28 GHz.
To generate wideband test signals with up to 2 GHz of modulation bandwidth at frequencies up to 44 GHz, the solution employs a precision arbitrary waveform generator (ARB) and vector signal generator, with wideband I/Q inputs, running signal creation software. Higher frequencies can be achieved through the use of an up-converter. This combination of hardware and software enables 5G candidate waveforms such as custom FBMC, OFDM and single-carrier to be generated. Integration of system-level design software with hardware further enables custom or proprietary algorithms and “what-if” scenarios to be evaluated, such as the coexistence of an LTE signal in the presence of an FBMC signal.
The testbed reference solution can also be used to demodulate and analyze test signals. In this case, 89600 VSA software is typically employed with the simulation software or on a number of different hardware options, including a signal analyzer, oscilloscope or PC that controls a variety of instruments or digitizers.
To illustrate the viability of this type of testbed reference solution, two test cases will be examined: a ~1 GHz wide custom OFDM signal at 28 GHz and a 2 GHz single-carrier signal at 73 GHz.
28 GHz Wideband OFDM Signal
For this case, the testbed reference solution combines a precision AWG with a vector signal generator with wideband I/Q inputs to produce wideband microwave test signals up to 44 GHz. Signal creation software enables the custom OFDM waveform to be created with approximately 1 GHz modulation bandwidth at 28 GHz (see Figure 2). Resource-mapping parameters were set for the preamble, pilot and data subcarriers, including the location and boosting of each resource block. I/Q values were set for the preamble, modulation and payload for pilot and data. The waveform is generated, then read into the AWG and played out using the AWG’s front panel software. The I/Q outputs of the AWG are fed into the wideband I/Q inputs on the vector PSG, and the PSG modulates the I/Q waveforms on a 28 GHz carrier. The test signal from the PSG RF output is analyzed using a 63 GHz high-performance oscilloscope with 89600 VSA software.
Figure 4 Example hardware setup for waveform generation and analysis at 73 GHz (mmWave amplifier and filter not shown).
The resulting test signal measurement with the 89600 VSA software (see Figure 3) comprises a six-trace display that shows (clockwise from upper left) the constellation, error vector magnitude (EVM) versus subcarrier, search time, OFDM equalizer channel frequency response, error summary and the ~1 GHz spectrum at the 28 GHz center frequency.
Figure 5 Integrating design software and test equipment to correct for linear amplitude and phase errors in the test signal (mmWave amplifier and filter not shown).
Figure 6 Demodulating a 73 GHz waveform with 2 GHz of modulation bandwidth. Constellation maximum used as the EVM normalization reference.
Single Carrier Signal at 73 GHz
For this case, the testbed reference solution configuration is extended to 73 GHz using a mmWave up-converter for signal generation and either a mmWave down-converter or waveguide smart mixer for signal analysis. The configuration shown in Figure 4 uses a microwave signal generator to provide the LO for the mmWave up-converter. A mmWave amplifier and filter at the up-converter output may be added to boost power and filter the spectrum. A waveguide smart mixer is used for signal analysis from 60 to 90 GHz, combined with a signal analyzer and oscilloscope. The waveguide smart mixer is connected to the output of the mmWave up-converter, and the IF output is fed into the signal analyzer for spectrum analysis. The auxiliary IF output is fed into the oscilloscope for wide bandwidth demodulation analysis with the VSA software.
At these frequencies and bandwidths, linear amplitude and phase errors may be caused within the signal chain by the AWG, vector signal generator, up-converter, waveguide smart mixer, cables/interconnects and signal analyzer. These were reduced by deriving the necessary vector corrections using the adaptive equalizer in the VSA software. The equalizer produces a complex-valued frequency response that can be used to minimize amplitude and phase errors. This is done by reading the frequency response into the system-level design software used to generate the wideband waveform, and then using it to pre-correct the waveform response (see Figure 5).
Figure 6 shows the demodulation analysis of the vector-corrected waveform at 73 GHz, with 2 GHz modulation bandwidth. Demodulating a 2 GHz wideband signal is typically quite difficult without adaptive equalization, due to hardware impairments across the wide bandwidth. However, in this example the linear amplitude and phase errors were corrected in simulation to generate a waveform that produced a low EVM without adaptive equalization.
Conclusion
The development of 5G includes an aggressive set of characteristics that will be difficult to achieve. A high degree of flexibility is needed to help researchers and engineers address these challenges and quickly respond to changes in direction as 5G evolves.
Test systems such as the 5G waveform generation and analysis reference solution combine hardware and software to create a flexible 5G waveform generation and analysis platform. This enables engineers and researchers to generate and analyze emerging 5G candidate waveforms. The software elements for the testbed provide flexibility in the types of 5G candidate waveforms being generated and analyzed. The hardware elements for the testbed provide flexibility and scalability from RF to microwave to mmWave and modulation bandwidths up to 2 GHz.
Greg Jue is an applications development engineer working on 5G applications at Keysight Technologies. He has worked in Keysight’s aerospace and defense applications team, the high performance oscilloscopes team and at EEsof, specializing in 802.11ac, LTE, WiMAX, aerospace and defense and software-defined radio applications. Jue wrote the design simulation section in Agilent Technologies’ LTE book and has authored numerous articles, presentations, application notes and whitepapers. Before joining HP/Agilent, he worked on the system design for the Deep Space Network at the Jet Propulsion Laboratory.
