by Rohde & Schwarz
Generating and analyzing candidate 6G waveforms above 100 GHz places demands on both the signal itself and the test equipment used to create it. The attraction of the sub-THz spectrum is straightforward: frequencies between 100 and 300 GHz offer extremely wide bandwidths and, in principle, the capacity needed for very high data rates. The difficulty is that the RF hardware becomes less forgiving as frequency rises. Phase noise, linearity, and power-amplifier efficiency all become more significant constraints, meaning that the choice of waveform and the measurement architecture are closely linked.
A Rohde & Schwarz testbed developed for sub-THz 6G research illustrates one way to address those problems using commercially available instruments. The system generates and analyzes DFT-s-OFDM waveforms and extends the measurement path into the 70 to 170 GHz region using external frequency-conversion front ends. A custom graphical interface coordinates the instruments, waveform parameters, and measurement sequence.
Why DFT-s-OFDM Matters at Sub-THz Frequencies
The waveform used in the testbed is based on discrete Fourier transform spread OFDM (DFT-s-OFDM). It combines many of the useful properties of OFDM with a lower peak-to-average power ratio. That distinction becomes particularly important at sub-THz frequencies, where available RF output power is limited and power-amplifier efficiency is a major concern. A waveform with a high PAPR forces an amplifier to operate with additional back-off to preserve linearity, reducing the usable transmitted power. Lowering PAPR therefore makes better use of the available RF hardware.
The proposed numerology is also substantially different from conventional 5G NR. The subcarrier spacing is 1.92 MHz, which is 128 times the 15 kHz reference spacing used in 5G NR. The larger spacing is intended to make the waveform more tolerant of the increased phase noise encountered at sub-THz frequencies while also allowing low-latency FFT processing. The cyclic prefix is approximately 79 ns, a value chosen to accommodate the short delay spreads expected in small-cell deployments.
The time-domain structure moves away from the traditional slot concept. Transmission can begin or end at nearly any DFT-s-OFDM symbol, producing a minimum latency of about 600 ns plus processing time. For control signaling and system information, the structure is organized around 1 ms subframes. Each subframe contains eight TDD periods, each with 208 symbols. A slightly extended cyclic prefix is inserted every 52nd symbol so that each 1 ms subframe contains an integer number of DFT-s-OFDM symbols. This avoids the much longer cyclic-prefix interval that would result from directly scaling the 5G NR structure.
Dynamic TDD provides additional flexibility. Three predefined downlink-to-uplink ratios are supported: 3:1, 1:1, and 1:3. Each 208-symbol TDD period includes a 12-symbol guard interval, while the uplink allocation changes depending on whether synchronization signal blocks or the physical random-access channel are present. The arrangement allows the waveform to adapt to different traffic patterns while retaining the short symbol timing required for low-latency operation.
Building the Signal Chain
The testbed separates the baseband/IF generation and analysis functions from the sub-THz frequency conversion. An R&S SFI100A vector signal generator produces the intermediate-frequency signal. That signal is applied to an external transmit frontend, either an FE110 or FE170, which translates it to the required RF frequency. The FE110 covers 70 to 110 GHz, while the FE170 extends operation from 110 to 170 GHz.
After upconversion, the RF signal travels through a waveguide section. A power amplifier and bandpass filter can be inserted in that path when additional gain or signal conditioning is required. At the receive side, a second frontend downconverts the sub-THz signal to an IF that can be analyzed by an R&S FSW signal and spectrum analyzer.
Synchronization is critical in a measurement system spanning several instruments and frequency converters. The SFI100A supplies a 10 MHz reference for frequency alignment, a 1 GHz reference split between the transmit and receive frontends, and a trigger to synchronize the start of the FSW measurement. Keeping the generation, conversion, and analysis chain on common references reduces the number of independent frequency and timing errors that would otherwise complicate evaluation of the waveform itself.
The architecture also makes it possible to work at IF without a front end when appropriate. In that mode, the signal generator’s frequency range used by the setup is 11 to 21 GHz. Adding the FE110 or FE170 moves the same basic measurement concept into the millimeter-wave and sub-THz bands without changing the overall workflow.
Controlling a Wideband 6G Experiment
A custom Python-based graphical interface was developed to manage the testbed. Rather than requiring the user to configure the signal generator, analyzer, front ends, and waveform software separately, the GUI provides a single control point for the entire chain. It handles instrument connections, waveform generation and loading, signal parameters, and measurement configuration.
The available controls illustrate the number of variables involved in a sub-THz waveform experiment. The user can select center frequency, output power, subcarrier spacing, uplink or downlink operation, TDD ratio, modulation, DMRS placement, FFT size, occupied bandwidth, and the number of TDD periods. The FFT size is fixed at 4096 in this implementation. Subcarrier spacing can be selected from 30 kHz to 1.92 MHz, and the number of active subcarriers determines occupied bandwidth in combination with that spacing.
Reference-signal configuration is equally important. The system can generate waveforms with synchronization signal blocks with Physical Random Access Channel (PRACH) or without either structure, depending on the measurement objective. Demodulation reference signals can be placed every 16th or 32nd symbol. The implementation also allows different pilot/data arrangements, enabling the engineer to examine how the receiver behaves with specific time-frequency structures rather than treating the signal as an undifferentiated block of occupied spectrum.
Once the parameters are entered, the software can generate a new waveform or load an existing waveform and its associated demodulation file. Applying the configuration sends the waveform and operating parameters to the signal generator and the demodulation information to the analyzer. For research work, this degree of automation is valuable because it makes repeated measurements with changed parameters more practical and reduces the chance that the generator and analyzer will be configured inconsistently.
Multi-Gigahertz Bandwidth in Practice
The example measurements in the application note show both downlink and uplink configurations. In each case, the generated signal occupies 3.84 GHz within a total configured bandwidth of 7.86 GHz. The downlink example includes an SSB, while the uplink example includes PRACH. The analyzer displays the captured waveform, power spectrum, result summary, and modulation constellation, allowing the engineer to examine both spectral behavior and demodulation quality in a single measurement.
Those bandwidth figures help put the test problem into perspective. At sub-THz frequencies, the challenge is not simply reaching a high carrier frequency. The measurement system must generate, translate, transport, receive, and analyze signals spanning several gigahertz while maintaining adequate frequency coherence and modulation quality. Every part of the chain, from references and converters to waveguide components and analyzer bandwidth, contributes to the measurement uncertainty.
The testbed does not represent a finalized 6G air interface; the waveform concept is intended for research and evaluation. What it demonstrates is the instrumentation architecture required to investigate candidate signals in frequency ranges that conventional microwave test setups cannot reach directly. DFT-s-OFDM provides a waveform with characteristics suited to power-limited sub-THz hardware, while external converters allow familiar vector signal generation and analysis techniques to be extended well beyond 100 GHz.
As 6G research pushes toward wider channels and higher carrier frequencies, the distinction between waveform design and RF test engineering will become increasingly difficult to separate. A promising modulation format still must survive real oscillators, amplifiers, converters, filters, and propagation paths. Sub-THz testbeds such as this one provide the bridge between a waveform that works in simulation and one that can be generated, transmitted, and measured with actual hardware.
Note: This article was adapted from Rohde & Schwarz Application Note 1SL417, “Sub-THz (6G) Communication Signal Generation and Analysis,” by Damjan Perencevic and Meik Kottkamp, February 2026.