DFT-s-OFDM - Uplink Waveform for 5G/6G MIMO
1. Introduction: Why do uplink waveforms matter?
Wireless communication systems have continuously evolved to support higher data rates, better coverage, improved spectral efficiency and lower power consumption. One important technology decision behind this evolution is the selection of the radio transmission waveform.
LTE introduced SC-FDMA (Single Carrier Frequency Division Multiple Access) for uplink transmissions, primarily to reduce the Peak-to-Average Power Ratio (PAPR) experienced by the User Equipment (UE). 5G New Radio (NR) retained this concept through DFT-spread OFDM (DFT-s-OFDM), while also supporting conventional CP-OFDM for uplink transmissions.
However, an interesting engineering challenge remains.
DFT-s-OFDM offers better power efficiency and lower PAPR, but its conventional NR implementation has limitations in spatial multiplexing. CP-OFDM, on the other hand, provides greater flexibility for multi-layer MIMO transmission but generally requires higher power-amplifier back-off.
As the industry moves towards 6G, researchers and 3GPP study groups are investigating how DFT-s-OFDM can be extended to support multi-layer uplink MIMO without losing its low-PAPR advantage.
This article explores the complete technical journey, starting from LTE SC-FDMA, explaining the DFT-s-OFDM transmitter and receiver architecture in 5G NR, comparing it with CP-OFDM, and examining why multi-layer DFT-s-OFDM is becoming an important 6G research direction.
It also connects waveform theory to practical protocol testing, UE power consumption, cell-edge performance, MIMO throughput, power-amplifier behaviour and real-world network deployment scenarios.
Before discussing DFT-s-OFDM, let us understand an important difference between uplink and downlink transmissions.
In a cellular network:
Downlink (DL): Data travels from the gNB/eNB to the UE.
Uplink (UL): Data travels from the UE to the gNB/eNB.
Both directions need to transmit information efficiently over limited radio spectrum. However, their hardware capabilities are quite different.
A base station has access to:
Higher transmission power.
Larger power amplifiers.
Multiple antenna elements.
Better cooling and power-supply capabilities.
More flexibility in RF hardware design.
A smartphone has very different constraints:
· Limited battery capacity.
· Small RF power amplifiers.
· Limited antenna space.
· Strict thermal constraints.
· Limited maximum transmission power.
For example, consider a smartphone transmitting uplink data to a 5G gNB.
The UE may be configured with 100 MHz bandwidth and 30 kHz subcarrier spacing. It might receive a PUSCH grant requiring transmission using a high-order modulation scheme.
The UE must generate the complete OFDM waveform, amplify it using its RF power amplifier and transmit it through its antenna.
Here is the problem: the instantaneous power of the transmitted waveform does not remain constant.
Some waveform samples have considerably higher amplitudes than others. If the power amplifier cannot reproduce these peaks accurately, the transmitted signal becomes distorted.
This distortion can cause:
· Increased EVM.
· Higher adjacent-channel leakage.
· Reduced achievable modulation order.
· Lower uplink throughput.
· Reduced transmission efficiency.
· Potential degradation of cell-edge coverage.
This is where DFT-s-OFDM becomes important.
The main objective of DFT-s-OFDM is to retain the advantages of OFDM while reducing the variations in the transmitted signal.
That makes it particularly attractive for battery-powered devices.
2. Understanding CP-OFDM
CP-OFDM stands for Cyclic Prefix Orthogonal Frequency Division Multiplexing.
It is one of the most widely used multicarrier transmission techniques in modern wireless systems.
LTE uses OFDM in downlink, while 5G NR uses CP-OFDM in both downlink and uplink.
2.1 How does CP-OFDM work?
The basic principle is to divide the available channel bandwidth into multiple narrowband orthogonal subcarriers.
Instead of transmitting all information using one high-speed carrier, the system distributes the information across multiple subcarriers.
Consider a simplified example:
A gNB allocates 12 subcarriers to a UE.
Each subcarrier carries one QPSK or QAM symbol during an OFDM symbol period.
The transmitter performs the following operations:
· Input binary data is generated.
· Bits are mapped into QPSK or QAM symbols.
· Symbols are mapped to their allocated frequency-domain subcarriers.
· IFFT converts the frequency-domain symbols into time-domain samples.
· Cyclic Prefix (CP) is added.
· The signal is converted into an analogue RF signal and transmitted.
The IFFT operation combines all subcarriers into one composite time-domain signal.
Mathematically:
Where:
2.2 Why is the cyclic prefix added?
Wireless channels introduce multipath propagation.
For example, when a UE transmits an uplink signal, the gNB may receive:
A direct signal.
A signal reflected from a building.
A signal reflected from a vehicle.
Other delayed signal components.
These delayed copies can interfere with the next OFDM symbol, causing Inter-Symbol Interference (ISI).
To reduce this problem, a portion of the end of the OFDM symbol is copied and placed at the beginning.
This is called the Cyclic Prefix.
image source: https://www.nrexplained.com/basics/ofdm
The cyclic prefix provides a guard interval and allows the receiver to perform frequency-domain equalization efficiently, provided the channel delay spread is within the cyclic-prefix duration.
However, CP does not eliminate every channel impairment. The receiver still needs synchronization, channel estimation and equalization.
2.3 The major limitation of CP-OFDM: PAPR
Now let us understand the most important limitation of CP-OFDM for uplink transmission.
Each subcarrier carries an individual modulated symbol. When all subcarriers are combined through the IFFT, their time-domain signals add together.
Sometimes the individual signals add constructively, producing a large amplitude peak.
At other times, they partially cancel each other.
Consequently, the transmitted signal has a varying envelope.
This leads to a high Peak-to-Average Power Ratio.
The PAPR is defined as:
Expressed in decibels:
Example:
Assume the average transmitted power is 100 mW, but the maximum instantaneous power reaches 1,000 mW.
This means the signal has a peak power 10 dB above its average power.
The UE power amplifier must handle these peaks without entering its nonlinear operating region.
To achieve this, the UE may need to reduce its average transmission power.
This reduction is known as Power Amplifier Back-Off (PBO).
And this is the fundamental reason why conventional OFDM is less attractive for power-limited uplink devices.
3. DFT-s-OFDM: Understanding the technology
Core concept
DFT-s-OFDM stands for Discrete Fourier Transform-spread Orthogonal Frequency Division Multiplexing.
It is also commonly called:
· DFT-spread OFDM.
· SC-FDMA in LTE terminology.
· Transform-precoded OFDM in 5G NR.
The important difference between CP-OFDM and DFT-s-OFDM is the additional DFT operation before subcarrier mapping.
In conventional CP-OFDM, modulated symbols are mapped directly to subcarriers and passed through an IFFT.
In DFT-s-OFDM, the modulated symbols first pass through a DFT operation. The resulting frequency-domain samples are then mapped to the allocated subcarriers before the IFFT operation.
This additional DFT operation spreads the information from each input symbol across the allocated subcarriers.
As a result, the generated time-domain signal has more single-carrier-like characteristics and generally exhibits lower PAPR.
3.1 Understanding the DFT spreading operation
Consider a UE that has received an uplink grant for 4 resource blocks.
Each resource block contains 12 subcarriers.
Therefore:
NRB=4, N RB_SC=12
The total number of allocated subcarriers is:
M=4 x 12 =48
Assume the UE has 48 QPSK symbols ready for transmission.
In CP-OFDM:
The 48 QPSK symbols are directly mapped to 48 subcarriers.
An IFFT generates the time-domain waveform.
A cyclic prefix is added.
In DFT-s-OFDM:
The 48 QPSK symbols are passed through a 48-point DFT.
The DFT generates 48 frequency-domain samples.
These samples are mapped to the 48 allocated subcarriers.
An IFFT generates the time-domain waveform.
A cyclic prefix is added.
The DFT operation is mathematically represented as:
Where:
x[n]= Input QAM symbols.
M = Number of input modulation symbols.
X[k]= DFT output.
k = Frequency-domain index.
The generated DFT samples are subsequently mapped to the allocated frequency-domain resources.
The receiver performs the reverse operations.
3.2 Why does DFT spreading reduce PAPR?
This is the most important concept to understand.
In CP-OFDM, each input QAM symbol is independently assigned to a subcarrier.
When multiple subcarriers combine in the time domain, their amplitudes can add constructively, producing high peaks.
In DFT-s-OFDM, the DFT operation spreads each input symbol across all the allocated subcarriers.
This changes the correlation and statistical characteristics of the transmitted time-domain waveform.
The resulting signal behaves more like a single-carrier transmission and generally has fewer extreme amplitude peaks.
Consider the following illustrative comparison:
Parameter | CP-OFDM | DFT-s-OFDM |
DFT spreading | No | Yes |
IFFT operation | Yes | Yes |
Cyclic prefix | Yes | Yes |
Signal envelope | Higher variations | Generally lower variations |
PAPR | Generally higher | Generally lower |
UE power efficiency | Lower in many PA-limited cases | Better potential |
Uplink coverage | May be power-limited | Can improve through reduced PA back-off |
MIMO flexibility | High | More restricted in conventional NR |
Important: DFT-s-OFDM does not produce a perfectly constant-envelope signal. Its PAPR depends on modulation, allocation size, spectrum mapping, filtering and other waveform parameters.
4. How does 5G NR implement DFT-s-OFDM?
5G NR continues to support the DFT-spread waveform for uplink transmission.
However, unlike LTE, 5G NR provides the network with the flexibility to select between two uplink waveforms:
CP-OFDM.
DFT-s-OFDM.
This is an important design improvement.
The network can use the waveform that better suits the UE's operating conditions and transmission requirements.
4.1 CP-OFDM versus DFT-s-OFDM in NR
The major difference lies in the transform precoding operation.
Processing stage | CP-OFDM PUSCH | DFT-s-OFDM PUSCH |
Channel coding | Yes | Yes |
Modulation | QPSK/QAM | QPSK/QAM and supported low-PAPR options |
Transform precoding | Disabled | Enabled |
Frequency resource mapping | After modulation | After DFT spreading |
IFFT | Yes | Yes |
CP insertion | Yes | Yes |
Receiver equalization | Frequency domain | Frequency domain |
Despreading | Not required | IDFT operation |
4.2 Transform precoding configuration
In 5G NR, transform precoding is associated with PUSCH transmission.
The relevant RRC configuration is:
transformPrecoder
The configuration is part of the PUSCH-related configuration framework in NR.
The UE also receives the necessary PUSCH resource allocation and transmission parameters through scheduling control information.
The key practical distinction is:
Transform precoding enabled: DFT-s-OFDM.
Transform precoding disabled: CP-OFDM.
The UE must support the applicable transmission configuration, and the network must configure and schedule it consistently.
4.3 How does the UE know which waveform to use?
The UE obtains its configuration through RRC signalling and the applicable PUSCH scheduling procedure.
The network configures the transform-precoding behaviour and the associated PUSCH parameters.
Depending on the configuration, DCI scheduling information can determine the relevant transmission resources and modulation parameters.
The protocol tester should inspect:
RRCReconfiguration.
PUSCH-Config.
Transform precoding configuration.
PUSCH grant.
DCI format 0_0 or 0_1.
PUSCH modulation and coding scheme.
Resource allocation.
DMRS configuration.
A common mistake during log analysis is to assume that every PUSCH transmission in NR uses CP-OFDM.
That is incorrect. NR supports both waveforms.
4.4 Important NR parameter: Transform precoding
The following is a simplified example of a PUSCH configuration:
PUSCH-Config
{
transformPrecoder enabled
resourceAllocation = type1
maxRank = 1
dmrs-UplinkConfig
frequencyHopping = disabled
}
This is illustrative, not a complete ASN.1 configuration.
The actual configuration is defined by the applicable 3GPP release and UE capabilities.
For DFT-s-OFDM, conventional NR PUSCH transmission has a single-layer restriction.
This is a major limitation when comparing it with CP-OFDM, which supports multi-layer uplink transmission.
4.5 Why does NR restrict DFT-s-OFDM to one layer?
The fundamental reason is to preserve its low-PAPR characteristics while keeping the transmitter and receiver processing manageable.
In a single-layer transmission, one DFT-spread symbol stream is generated and transmitted.
In a multi-layer MIMO transmission, multiple independent streams must be mapped to different spatial dimensions.
When these streams are combined using spatial precoding, their signals may create additional amplitude variations.
Consequently, the PAPR advantage of conventional DFT-s-OFDM can be reduced.
This is one of the problems that 6G research is attempting to solve.
5. DFT-s-OFDM versus CP-OFDM: Detailed technical comparison
Let us compare both waveforms from the perspective of a wireless system architect.
Technical parameter | CP-OFDM | DFT-s-OFDM |
Fundamental technology | Multicarrier OFDM | DFT-spread OFDM |
DFT spreading | No | Yes |
IFFT | Yes | Yes |
PAPR | Higher generally | Lower generally |
Power amplifier back-off | Higher generally | Lower generally |
UE power consumption | Potentially higher | Potentially lower |
Spectral efficiency | High | High, with mapping constraints |
Resource allocation flexibility | High | More restricted |
Non-contiguous allocation | Supported | Restricted in conventional NR |
Uplink MIMO | Multiple layers | One layer in conventional NR |
Receiver processing | FFT and equalization | FFT, equalization and IDFT |
Cell-edge potential | Depends on PA limitations | Can benefit from lower PAPR |
High-order modulation | Supported | Supported, subject to configuration |
Typical LTE uplink | Not the selected PUSCH waveform | SC-FDMA |
Typical NR uplink | Supported | Supported |
5.1 Power amplifier efficiency
The most important benefit of DFT-s-OFDM is its potential to improve power amplifier efficiency.
A power amplifier has a relatively linear operating region.
When the input signal exceeds this region, the output signal may become distorted.
This can produce:
· AM-AM distortion.
· AM-PM distortion.
· EVM degradation.
· Adjacent Channel Leakage Ratio (ACLR) degradation.
Image source: https://www.mdpi.com/2079-9292/10/9/1073
To avoid this, the transmitter operates with sufficient back-off.
For a waveform with high PAPR, the difference between average and peak power is larger.
For a lower-PAPR waveform, the UE may operate closer to the amplifier's efficient region while still satisfying the applicable RF requirements.
The practical advantage is not simply lower PAPR. It is the possibility of transmitting more useful average RF power within the same peak-power and distortion constraints.
5.2 Uplink coverage improvement
Consider two smartphones connected to the same gNB.
Both have a maximum permitted transmit power of 23 dBm.
Assume the CP-OFDM transmission needs 7 dB of PA back-off, while a particular DFT-s-OFDM configuration needs 4 dB.
These are hypothetical values used to demonstrate the principle, not universal 3GPP requirements.
Parameter | CP-OFDM | DFT-s-OFDM |
Maximum available power | 23 dBm | 23 dBm |
Assumed PA back-off | 7 dB | 4 dB |
Illustrative average operating power | 16 dBm | 19 dBm |
The difference is:
A 3 dB difference corresponds to approximately twice the RF power.
What does this mean in a real network?
The UE transmitting using DFT-s-OFDM may have more uplink power headroom under the assumed operating conditions.
This can improve the received signal quality at the gNB, potentially enabling:
Higher PUSCH reliability.
Better uplink coverage.
Improved cell-edge throughput.
Reduced retransmissions.
Better battery efficiency for a given uplink data requirement.
However, the actual coverage gain depends on path loss, receiver noise, interference, MCS, antenna configuration and power-control limits.
6. Practical example: How waveform selection affects 5G uplink throughput
Let us take a realistic 5G NR test scenario.
Example: 5G NR uplink throughput test
Channel bandwidth: 100 MHz
Subcarrier spacing:30 kHz
Duplex mode:TDD
UE transmit power:23 dBm
Modulation:64-QAM
PUSCH allocation:80 MHz equivalent
Illustrative test configuration; actual RB count depends on the configured carrier and guard bands.
Suppose we conduct two uplink tests.
Test 1: CP-OFDM
UE is configured for CP-OFDM.
PUSCH is scheduled.
UE transmits using its available power.
The power amplifier experiences higher peak variations.
The UE requires additional PA back-off to meet the RF quality requirements.
Test 2: DFT-s-OFDM
· UE is configured for DFT-s-OFDM.
· PUSCH is scheduled using transform precoding.
· DFT spreading is performed before IFFT.
· The transmitted waveform exhibits lower PAPR.
· The UE may achieve better average RF output power under the same peak and distortion constraints.
Now compare the possible test observations:
KPI | CP-OFDM | DFT-s-OFDM |
PAPR | Higher | Lower |
Average RF output power | May be lower | May be higher |
PA efficiency | May be lower | May be higher |
EVM | Depends on PA operating point | Depends on PA operating point |
PUSCH BLER | Channel dependent | Channel dependent |
UL throughput | May be power-limited | May improve if power-limited |
The important point is that a lower PAPR does not automatically guarantee higher throughput.
Throughput is affected by several other factors, including:
MCS selection.
Number of allocated RBs.
PUSCH symbols.
TDD UL/DL pattern.
DMRS overhead.
HARQ retransmissions.
Channel quality.
Power-control configuration.
Engineering takeaway: DFT-s-OFDM provides a potential physical-layer power advantage. The benefit becomes visible when the UE is actually limited by its RF power-amplifier operating conditions.
7. Why is MIMO important for 6G uplink?
Now we reach the most interesting part of the article.
Why does 6G want DFT-s-OFDM for multi-layer MIMO transmission?
To answer this, we must first understand the role of MIMO in wireless communication.
MIMO stands for Multiple Input Multiple Output.
It uses multiple transmitting and receiving antennas to improve wireless communication performance.
In a 5G NR uplink scenario, the UE can transmit using multiple antenna ports, and the gNB can receive signals using multiple antennas.
The system can exploit:
Spatial multiplexing.
Transmit diversity.
Receive diversity.
Beamforming.
Spatial channel separation.
7.1 Single-layer transmission
Consider a smartphone transmitting data to a 5G gNB.
The UE has two transmit antenna branches, but the scheduled PUSCH transmission uses one spatial layer.
Both antenna branches can participate in transmitting the same layer through an applicable precoding arrangement.
The important point is that two transmit antennas do not necessarily mean two independent data streams.
The UE may use multiple antenna ports to transmit one layer.
Conventional NR DFT-s-OFDM supports this single-layer transmission approach.
7.2 Multi-layer transmission
Now consider a UE transmitting two independent data streams simultaneously.
For example:
Layer 1 carries one portion of the user's data.
Layer 2 carries another portion.
The gNB receives both streams and separates them using MIMO signal processing.
The advantage is higher spectral efficiency.
In an idealized scenario, two independent spatial layers can deliver approximately twice the data rate of one layer using the same time-frequency resources.
However, this requires suitable channel conditions.
The gNB must be able to distinguish the spatial layers.
For example, a 2-layer transmission requires a sufficiently well-conditioned MIMO channel, appropriate UE capabilities, suitable precoding and adequate signal-to-interference-plus-noise ratio.
7.3 Why is multi-layer MIMO important in 6G?
6G aims to support significantly higher uplink data rates and more demanding applications.
Consider the following use cases:
1. Extended Reality (XR)
An XR headset may continuously upload camera feeds, spatial information, user movements and sensor data.
Multiple spatial layers can increase the uplink capacity.
2. AI-powered mobile devices
Future AI-enabled devices may upload large amounts of sensor information, image data and intermediate computation results to edge AI servers.
3. Industrial automation
Robots and automated production equipment may transmit multiple high-resolution camera streams and real-time sensor information.
4. High-resolution live video
A mobile device streaming 8K video or multiple camera feeds can benefit from higher uplink throughput.
5. Connected vehicles
Vehicles may transmit multiple camera, radar-processing and sensor data streams to infrastructure and edge computing systems.
In these scenarios, relying on a single spatial layer can become a throughput limitation.
This creates the motivation for combining DFT-s-OFDM with multi-layer MIMO.
8. The challenge: Why is DFT-s-OFDM difficult to extend to MIMO?
The difficulty is not simply enabling two transmit antennas.
The real challenge is maintaining the low-PAPR properties while transmitting multiple independent spatial layers.
Let us understand this technically.
8.1 Single-layer DFT-s-OFDM
In a single-layer DFT-s-OFDM transmission:
x[n] -> DFT -> Subcarrier Mapping -> IFFTOne modulated symbol stream is spread through the DFT operation.
The resulting waveform has lower PAPR characteristics compared with conventional CP-OFDM.
8.2 Multi-layer transmission
In a two-layer transmission, two independent symbol streams are generated.
Each stream can undergo DFT spreading.
However, the streams must subsequently be mapped and transmitted through the available antenna ports.
The spatial precoder combines the layers.
A simplified MIMO transmission model is:
Where:
s = Spatial-layer symbol vector.
W = Transmit precoding matrix.
H = MIMO wireless channel.
y = Received signal.
n = Noise and interference.
For two layers:
The precoder combines the two layers across transmit antenna ports.
For example, a simplified two-transmit-antenna arrangement can be represented as:
This is an illustrative unitary precoding matrix, not a statement about a mandatory NR or 6G precoder.
The signals transmitted from the antenna ports are combinations of the two spatial layers.
8.3 Why can PAPR increase?
Consider two independent waveforms.
Even if both have relatively low PAPR individually, their combination can produce higher instantaneous peaks.
The signal envelope depends on:
Relative signal amplitudes.
Relative phases.
Precoding coefficients.
Layer correlation.
Modulation symbols.
Frequency-domain mapping.
The precoding operation can therefore weaken the low-PAPR characteristics of the original DFT-spread signals.
This is the central engineering trade-off:
The 6G waveform design challenge
Low PAPR and better PA efficiency
Preserve the main advantage of DFT-s-OFDM.
Multiple spatial layers
Increase uplink throughput using MIMO.
Controlled PAPR degradation
Ensure the additional layers do not eliminate the power-efficiency advantage.
The 6G research objective is to find a practical balance between these requirements.
9. How does 6G propose to support multi-layer DFT-s-OFDM?
6G research is investigating different approaches to preserve the low-PAPR properties of DFT-s-OFDM while allowing multiple spatial layers.
The objective is not simply to replace CP-OFDM.
Instead, the idea is to provide the network with more waveform flexibility, so it can select an appropriate transmission technique depending on the UE's coverage, channel quality and throughput requirements.
The following are important research directions.
9.1 Rank-2 DFT-s-OFDM
One of the interesting directions is to extend DFT-s-OFDM from conventional single-layer transmission to two spatial layers.
This is generally referred to as Rank-2 DFT-s-OFDM.
Consider a UE with two transmit antenna branches.
In a conventional single-layer arrangement:
One transport data stream is processed.
DFT spreading is performed.
The resulting waveform is mapped and transmitted.
In a Rank-2 arrangement:
Two spatial layers are generated.
Each layer carries independent information.
DFT-spreading and spatial processing are performed according to the selected waveform design.
The two layers are transmitted through the available antenna ports.
The gNB separates the two layers using MIMO receiver processing.
The benefit is that the UE can increase its uplink throughput without completely abandoning the low-PAPR waveform approach.
Why start with two layers instead of immediately supporting four or eight?
Because increasing the number of layers introduces additional challenges:
Higher peak-power variations.
More complicated precoding.
Greater receiver complexity.
More demanding channel estimation.
Increased RF implementation complexity.
Two layers can provide a useful balance between throughput and power efficiency.
However, Rank-2 DFT-s-OFDM is a 6G research direction, not a feature that should be assumed to be available in current 5G NR UE implementations.
9.2 Multi-layer transmission with different precoding approaches
Another research direction involves investigating different spatial precoding matrices.
In conventional single-layer DFT-s-OFDM, the spatial processing is relatively straightforward.
For multi-layer transmission, the precoder becomes an important design variable.
Consider a two-layer transmission.
The precoding matrix determines how the two spatial streams are distributed across the transmit antenna ports.
Different precoding arrangements can produce different waveform envelopes.
For example:
Precoding approach | Expected engineering consideration |
Identity-based mapping | Preserves a relatively straightforward layer-to-port relationship |
Conventional linear precoding | Can improve spatial transmission but affect PAPR |
Alternative layer-mixing precoding | May improve spatial flexibility, with PAPR trade-offs |
Optimized PAPR-aware precoding | Attempts to balance spatial gain and peak power |
9.3 Frequency-domain spectrum shaping
Another interesting direction is Frequency Domain Spectrum Shaping (FDSS).
The basic idea is to modify the frequency-domain samples after DFT spreading to control the transmitted waveform's envelope.
The processing can be represented as:
Where:
X[k] = DFT output.
W[k] = Frequency-domain shaping coefficient.
X'[k] = Shaped frequency-domain output.
The shaped samples are subsequently mapped to the allocated resources.
The objective is to reduce signal peaks while controlling unwanted spectral emissions.
The shaping coefficients must be designed carefully because excessive spectral shaping can introduce:
In-band distortion.
Receiver equalization complexity.
Changes in the effective noise characteristics.
Spectral-efficiency penalties.
FDSS is therefore a trade-off between PAPR reduction and signal quality.
9.4 π/2-BPSK modulation
Another important technique associated with low-PAPR uplink transmission is π/2-BPSK.
In conventional BPSK, the modulation symbols are represented by two opposite constellation points.
In π/2-BPSK, the constellation is rotated by 90 degrees between successive modulation symbols.
The rotation can be represented as:
where,
This alternating phase rotation helps shape the time-domain signal and can reduce PAPR.
Why is this useful?
A UE transmitting with a low-order modulation scheme is often operating in a coverage-limited environment.
π/2-BPSK can provide favourable power-amplifier operating characteristics for these conditions.
It is particularly interesting for:
Cell-edge transmission.
Low-SINR conditions.
Coverage extension.
Power-limited terminals.
It should be noted that π/2-BPSK alone does not solve the multi-layer MIMO challenge. It is one component of the broader low-PAPR waveform design.
9.5 Frequency-domain truncation
Frequency-domain truncation is another technique being investigated for 6G.
The idea is to modify the frequency-domain representation to reduce the peak variations of the transmitted signal.
A simplified representation is:
Here, Kactive represents the retained frequency components.
This technique changes the spectral representation and can influence the resulting time-domain envelope.
However, truncation needs to be designed together with resource allocation, spectral confinement and receiver processing.
A 2026 Qualcomm RAN1 contribution discusses π/2-BPSK and frequency-domain truncation as potential low-PAPR waveform enhancements for 6G uplink. It also proposes studying multi-layer DFT-s-OFDM.
9.6 Dynamic waveform switching
This could become one of the most useful capabilities for future radio networks.
Instead of permanently using one waveform, the network could select between CP-OFDM and DFT-s-OFDM depending on transmission conditions.
For example:
UE operating condition | Potential waveform preference |
Cell-edge UE | DFT-s-OFDM |
UE with limited transmit power | DFT-s-OFDM |
High-throughput UE with good channel conditions | CP-OFDM or multi-layer DFT-s-OFDM |
UE requiring multiple spatial layers | Multi-layer waveform with suitable MIMO support |
Battery-sensitive device | Low-PAPR waveform, subject to implementation |
High mobility | Selected based on channel and implementation performance |
This allows the network to balance power efficiency and spectral efficiency.
10. 6G MIMO waveform example: A practical comparison
Let us consider a practical example to understand the difference between single-layer and multi-layer transmission.
Assume a future 6G UE has two transmit antenna branches and is connected to a base station with a larger antenna array.
The UE needs to upload high-resolution video to an edge computing server.
We compare three possible transmission arrangements.
Scenario A: Single-layer DFT-s-OFDM
The UE uses one spatial layer.
The transmitter performs DFT spreading and generates a low-PAPR waveform.
Advantages:
· Lower PAPR.
· Better PA operating efficiency potential.
· Good coverage characteristics.
· Reduced RF power requirement for a given transmission quality.
Limitation:
· Only one independent spatial stream is transmitted.
· Therefore, the achievable throughput may be limited.
Scenario B: Two-layer CP-OFDM
The UE uses two independent spatial layers with CP-OFDM.
Advantages:
· Higher spatial multiplexing capability.
· Increased potential throughput.
· Flexible resource mapping.
· Mature multi-layer MIMO processing.
Limitation:
Higher PAPR can increase the power-amplifier back-off requirement.
This may reduce average transmitted power, particularly for power-limited UEs.
Scenario C: Two-layer DFT-s-OFDM
The UE uses two spatial layers with a multi-layer DFT-s-OFDM design.
Advantages:
Spatial multiplexing gain.
Potentially lower PAPR than a comparable CP-OFDM arrangement.
Better potential PA efficiency.
Opportunity to improve uplink throughput and coverage together.
Limitation:
The transmitter and receiver need additional waveform and MIMO processing.
The low-PAPR benefit also needs to be verified for the actual layer mapping and precoding arrangement.
10.1 Comparison of the three approaches
Feature | Single-layer DFT-s-OFDM | Two-layer CP-OFDM | Two-layer DFT-s-OFDM |
Spatial layers | 1 | 2 | 2 |
PAPR | Low generally | Higher generally | Design dependent |
PA efficiency | Good potential | More challenging | Good potential |
Spatial multiplexing | Limited | Yes | Yes |
Peak throughput | Limited by one layer | Higher potential | Higher potential |
Receiver complexity | Lower | Established MIMO processing | Additional processing |
5G NR availability | Yes | Yes | Not supported as conventional NR PUSCH |
10.2 Illustrative throughput calculation
Assume the following simplified values:
Allocated bandwidth: 100 MHz.
Effective spectral efficiency per layer: 3 bits/s/Hz.
One spatial layer for DFT-s-OFDM.
Two spatial layers for multi-layer transmission.
For one layer:
R1= 100 x 3 = 300Mbps
For two layers:
R2 = 100x 3 x 2 = 600Mbps
The theoretical rate doubles in this simplified example.
But actual throughput will be lower because of:
DMRS overhead.
Coding and scheduling.
Control-channel overhead.
TDD UL/DL allocation.
Guard periods.
HARQ retransmissions.
Channel quality.
Implementation losses.
The important observation is that multi-layer DFT-s-OFDM could offer a way to combine spatial multiplexing with a low-PAPR waveform.
It is not guaranteed to achieve twice the throughput or the same PAPR as single-layer DFT-s-OFDM.
11. Real-world use cases: Where can 6G DFT-s-OFDM MIMO help?
Let us connect the waveform design to practical network deployments.
11.1 Smartphone uplink at the cell edge
Consider a user located several kilometres from a cellular base station.
The downlink signal may still be acceptable because the base station has a powerful transmitter.
However, the uplink becomes challenging because the smartphone has limited transmit power.
The UE reaches its maximum transmit power, and the gNB starts observing poor uplink SINR.
The network may respond by reducing the MCS or allocating more resources.
This reduces throughput.
A lower-PAPR waveform could allow the UE's power amplifier to operate more efficiently.
The potential benefit is improved uplink link budget and cell-edge data performance.
6G opportunity: Use low-PAPR DFT-s-OFDM when the UE is power-limited and enable multiple layers when channel conditions permit.
11.2 XR and immersive communication
Future XR devices may continuously upload:
Camera streams.
Environmental information.
Motion and position data.
Hand and eye-tracking information.
Spatial mapping information.
These applications require a combination of high uplink throughput, low latency and energy efficiency.
A single layer may not provide sufficient throughput in demanding scenarios.
Multi-layer DFT-s-OFDM could help increase uplink capacity while retaining a lower-PAPR transmission architecture.
11.3 Industrial robotics
In a smart factory, multiple robots may continuously transmit video and sensor data to edge computing systems.
The network needs to support:
Reliable data transmission.
Predictable latency.
High uplink capacity.
Efficient power consumption.
For battery-operated robots, reducing the RF transmission energy can be particularly useful.
Multi-layer uplink transmission could improve the capacity available to devices with suitable MIMO channels.
However, deterministic latency and reliability require more than waveform selection. Scheduling, QoS, channel coding and network architecture remain essential.
11.4 Non-Terrestrial Networks (NTN)
Satellite communication introduces additional link-budget challenges.
The UE may be transmitting to a satellite or an NTN gateway through a satellite radio link.
The communication system can experience:
· Large propagation losses.
· Long propagation delays.
· Doppler shifts.
· Limited terminal transmit power.
· Restricted antenna size.
For such scenarios, a power-efficient waveform is attractive.
DFT-s-OFDM research is relevant because lower PAPR can help improve power-amplifier operation.
This is particularly interesting for handheld satellite terminals and low-power connected devices.
However, NTN also requires dedicated timing, frequency and Doppler compensation. DFT-s-OFDM does not independently solve those problems.
11.5 AI-enabled mobile devices
Future smartphones may execute AI workloads locally and offload computationally intensive operations to edge servers.
For example:
A mobile device captures a high-resolution video, performs initial AI processing and uploads selected image frames and feature information to an edge AI platform.
The uplink needs to support both:
High data rate.
Efficient energy consumption.
Multi-layer DFT-s-OFDM could be useful when the device needs higher throughput but still operates under battery and RF power constraints.
12. What should we validate?
This section is particularly important for engineers working on UE, gNB and RAN protocol verification.
Understanding the waveform is not sufficient. A protocol tester should be able to identify the waveform selection, verify the associated configuration and investigate its RF impact.
12.1 5G NR DFT-s-OFDM test cases
In NR, the following parameters are important.
Parameter | Why should you verify it? |
transformPrecoder | Determines the transform-precoding configuration |
resourceAllocation | Defines PUSCH resource allocation |
frequencyHopping | Affects frequency resource mapping |
dmrs-UplinkConfig | Determines DMRS configuration |
mcs-Table | Determines supported modulation and coding selection |
maxRank / applicable layer configuration | Determines the supported transmission rank |
p0-PUSCH-Alpha | Participates in PUSCH power control |
tpc-Accumulation | Controls accumulated transmit power adjustments |
The exact applicability of individual parameters depends on the RRC configuration and 3GPP release.
12.2 How to identify waveform selection in UE logs
A practical debugging sequence is:
1. Step 1: Check RRC configuration
· Decode the RRCReconfiguration message and identify the PUSCH configuration.
· Check whether transform precoding is enabled or disabled.
2. Step 2: Check the PUSCH grant
· Decode DCI format 0_0 or 0_1.
Verify:
· Frequency-domain resource allocation.
· Time-domain resource allocation.
· MCS.
· HARQ process number.
· New Data Indicator.
· TPC command.
3. Step 3: Check the UE transmission
Inspect the PUSCH transmission in the UE PHY log.
Confirm that the transmitter is using the expected waveform processing.
4. Step 4: Check RF measurements
Measure:
· PUSCH EVM.
· ACLR.
· PAPR.
· Average transmit power.
· Peak power.
· Power-amplifier operating point.
5. Step 5: Correlate with throughput
Compare the following:
PUSCH BLER.
HARQ retransmissions.
UL SINR.
MCS.
Achieved throughput.
UE power headroom.
12.3 Important troubleshooting example
Suppose the UE is configured for DFT-s-OFDM.
You observe:
KPI | Observation |
UE transmit power | Near maximum |
PUSCH SINR | Low |
PUSCH BLER | 12% |
MCS | 16QAM |
UL throughput | Lower than expected |
PHR | Low or negative headroom |
What would you investigate?
First, verify whether the UE is actually operating with transform precoding.
Next, examine the PUSCH power-control parameters and the maximum UE transmission power.
Then inspect:
Path loss estimation.
Open-loop power-control configuration.
Closed-loop TPC commands.
PUSCH allocation size.
DMRS quality.
Interference.
RF EVM and ACLR.
If the UE is already transmitting at its maximum power, changing the waveform alone may not solve the problem.
The advantage of DFT-s-OFDM becomes meaningful when PA back-off is an important limitation.
12.4 Additional 6G test cases for multi-layer DFT-s-OFDM
For future 6G validation, the test plan should include:
Test case | Expected validation |
Rank-1 DFT-s-OFDM | Baseline PAPR and throughput |
Rank-2 DFT-s-OFDM | Multi-layer transmission functionality |
PAPR measurement | Compare against equivalent CP-OFDM |
Precoder switching | Validate waveform quality across precoders |
Power amplifier testing | Verify EVM, ACLR and efficiency |
Cell-edge performance | Measure uplink BLER and throughput |
Mobility | Evaluate waveform performance under Doppler |
Layer imbalance | Verify performance when layer powers differ |
High-order modulation | Evaluate EVM and BLER |
Dynamic waveform switching | Validate waveform changes and scheduling consistency |
For a proper comparison, the CP-OFDM and DFT-s-OFDM tests should use equivalent:
Bandwidth.
Resource allocation.
Modulation and coding.
Output-power constraints.
Channel conditions.
Antenna configuration.
Otherwise, the observed performance differences may not be caused by the waveform itself.
13. Final engineering perspective: Where is uplink waveform technology heading?
Evolution from LTE to 6G
Generation | Uplink waveform approach | Main objective |
LTE | SC-FDMA | Reduce PAPR and improve UE power efficiency |
5G NR | CP-OFDM and DFT-s-OFDM | Balance flexibility, throughput and power efficiency |
6G research | Enhanced DFT-s-OFDM and multi-layer MIMO | Combine low PAPR with higher spatial throughput |
The three major engineering objectives
1. Improve power efficiency
The UE should be able to transmit useful RF power without requiring excessive power-amplifier back-off.
DFT-s-OFDM provides an attractive starting point.
2. Increase uplink spectral efficiency
Future devices will generate much more uplink data.
Single-layer transmission may not be sufficient for demanding applications.
Multi-layer MIMO can increase the amount of information transmitted using the same time-frequency resources.
3. Maintain waveform flexibility
There will not necessarily be one waveform that is optimal for every scenario.
A cell-edge smartphone, an XR headset, an industrial robot and a satellite-connected device have different requirements.
The network should be able to select a suitable waveform based on the operating environment.
Conclusion
DFT-s-OFDM is an excellent example of how a technology introduced to solve a practical hardware limitation in LTE can become relevant again in the development of 6G.
LTE introduced SC-FDMA to address the high PAPR problem associated with uplink OFDM transmission from power-constrained mobile devices.
5G NR retained the same underlying DFT-spreading principle but introduced the flexibility of supporting both CP-OFDM and DFT-s-OFDM.
Now, 6G research is attempting to address the next limitation: how to combine the power-efficiency advantages of DFT-s-OFDM with the spatial multiplexing benefits of MIMO.
The real goal of 6G is not simply to transmit more bits per second. It is to transmit more bits per second using the available spectrum and device power more intelligently.
That is why DFT-s-OFDM deserves attention from every engineer working in LTE, 5G NR and future 6G radio technologies.
References
Specification / Reference | Relevant topic |
3GPP TS 36.211 | LTE physical channels and SC-FDMA signal generation, particularly clause 5.6 |
3GPP TS 36.212 | LTE multiplexing and channel coding |
3GPP TS 36.213 | LTE physical-layer procedures |
3GPP TS 38.211 | NR physical channels and modulation, including PUSCH waveform generation |
3GPP TS 38.212 | NR multiplexing and channel coding |
3GPP TS 38.213 | NR physical-layer procedures |
3GPP TS 38.214 | NR physical-layer data procedures |
3GPP TS 38.331 | NR RRC protocol and PUSCH configuration |
3GPP TR 38.914 | 6G scenarios and requirements |
3GPP RAN1 R1-2600366 | Tejas Networks contribution on 6G waveform design |
3GPP RAN1 R1-2601268 | Qualcomm contribution on 6G uplink waveforms |
External references:
AI Usage Disclosure
Artificial Intelligence (AI) tools were used to assist in preparing and organizing this technical article. The technical concepts, explanations, illustrative examples and engineering discussions are intended for educational purposes. Readers should refer to the relevant 3GPP specifications and official technical documents for normative requirements and implementation details.
Comments