⚡ electrical engineering

Cascaded SLM-PTS and Joint Spatial-Temporal Phase Optimization with POCS for Low-PAPR Hybrid OFDM-SC-FDMA in MIMO 6G Uplinks

This paper proposes a novel transmitter architecture for 6G MIMO uplinks that integrates a hybrid OFDM-SC-FDMA waveform with cascaded SLM-PTS, joint spatial-temporal phase optimization, and POCS refinement to achieve a significant 4.8 dB reduction in PAPR while maintaining competitive BER and EVM performance at the cost of increased processing complexity.

Original authors: KOKA SAI SRIKANTH, KOLACHALAMALA RAJA RUTHVIKA, KATHI SAI GANESH REDDY, GANGULA VISHWAPRIYA REDDY, GOLLA TEJA SAI KIRAN, MEKALA SPOORTHI

Published 2026-10-05

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: KOKA SAI SRIKANTH, KOLACHALAMALA RAJA RUTHVIKA, KATHI SAI GANESH REDDY, GANGULA VISHWAPRIYA REDDY, GOLLA TEJA SAI KIRAN, MEKALA SPOORTHI

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the invisible rivers of data that flow between our phones and cell towers, a hidden battle is fought over power. The signals that carry our calls and videos are built by stacking many tiny waves on top of one another, a technique that allows for fast and clear communication. However, this stacking creates a problem: occasionally, these waves align perfectly to create a massive, sudden spike in energy. This spike, known as a peak-to-average power ratio, forces the battery-powered transmitters in our devices to either waste energy by operating inefficiently or to distort the signal, causing errors. For the next generation of wireless networks, where devices must be smaller, faster, and more energy-efficient, taming these spikes is a critical engineering challenge.

A team of researchers from Koneru Lakshmaiah Education Foundation in India has proposed a new way to smooth out these signals before they are sent. Their work focuses on a hybrid system that blends two different methods of sending data, then applies a series of clever filters to flatten the energy spikes. Instead of trying to solve the problem with a single, heavy-handed tool, they built a layered approach. First, they mix the data in a way that naturally reduces the likelihood of extreme spikes. Then, they generate several slightly different versions of the signal, each with its energy peaks shifted to different places, and select the smoothest one. Next, they break the signal into smaller chunks and adjust the timing of each chunk to further lower the highest peaks across all the antennas on a device simultaneously. Finally, they apply a mathematical refinement that gently clips off any remaining sharp edges, introducing only small, in-band residual distortion without significantly damaging the information inside.

The results of their computer simulations are striking. In a test environment mimicking a modern wireless network with multiple antennas, the standard method of sending data produced signal spikes that were roughly 10.2 decibels high. The new method reduced this to about 5.4 decibels. This difference of nearly 5 decibels is significant; it translates to a potential increase in the efficiency of the power amplifier, the component that boosts the signal for transmission, by a factor of nearly 1.7. In practical terms, this means a device could transmit the same amount of data using much less battery power, or transmit further with the same battery.

The researchers compared their method against other existing techniques that try to solve the same problem. Some older methods simply cut off the top of the signal spikes, which is like trimming the peaks of a mountain. While this lowers the height, it leaves behind a jagged landscape that distorts the message, leading to a point where increasing the signal strength no longer improves the connection quality. The new method avoids this severe error floor entirely. While it does leave behind small amounts of in-band distortion, this residual effect is negligible compared to the noise in typical signal ranges. In tests, it maintained the same clarity and error rate as the best standard methods, while other techniques showed a noticeable drop in performance. The only cost for this improvement is the time the device's processor needs to calculate the best signal shape. The simulations showed that this calculation takes significantly longer than standard methods, suggesting that the technology is currently best suited for scenarios where processing time is less critical than battery life, or for future hardware that can handle the extra work.

The team also explored how to manage the extra information needed to tell the receiver how the signal was adjusted. They found that by carefully coordinating the adjustments across different antennas and over time, they could keep the amount of extra data required very small. This ensures that the efficiency gained in power is not lost by wasting bandwidth on control messages. The study concludes that while the computational cost is high, the trade-off is worth it for the substantial gains in power efficiency and signal quality. As wireless networks evolve toward the sixth generation, this layered approach offers a promising path to keeping our devices connected longer and more reliably, turning a fundamental physical limitation into a manageable engineering task.

Technical Summary: Cascaded SLM-PTS and Joint Spatial-Temporal Phase Optimization with POCS for Low-PAPR Hybrid OFDM-SC-FDMA in MIMO 6G Uplinks

Problem Statement
High Peak-to-Average Power Ratio (PAPR) remains a critical bottleneck for power-limited uplink terminals in sixth-generation (6G) networks. In Orthogonal Frequency-Division Multiplexing (OFDM), the superposition of independent subcarriers creates a heavy-tailed envelope distribution, forcing Power Amplifiers (PA) to operate with large back-offs to avoid compression and spectral regrowth. This reduces efficiency and battery life. While Single-Carrier FDMA (SC-FDMA) offers lower PAPR, it sacrifices spectral flexibility. Furthermore, in Multiple-Input Multiple-Output (MIMO) systems, the figure of merit is the maximum PAPR across all transmit antennas, as each branch requires a linear PA. Existing solutions often trade PAPR reduction against in-band distortion (clipping), out-of-band (OOB) emissions, side information overhead, or computational complexity.

Methodology
The paper proposes a layered transmitter architecture designed to address these trade-offs through a combination of waveform shaping, probabilistic phase rotation, joint optimization, and convex projection. The system operates on a hybrid OFDM/SC-FDMA waveform where a fraction $\rho$ of data symbols is DFT-spread. The processing chain consists of four distinct stages:

Hybrid Waveform Mapping: A fraction of subcarriers are DFT-spread (SC-FDMA-like) while the remainder remain standard OFDM. This provides a baseline PAPR reduction function of the spreading ratio $\rho$ .
Cascaded SLM-PTS: A Selected Mapping (SLM) stage is applied first to generate $U$ phase-rotated candidates, selecting the one with the lowest PAPR. This output is then fed into a Partial Transmit Sequence (PTS) stage, where the signal is partitioned into $V$ sub-blocks and optimized with phase factors from a set of size $W$ . The cascade is ordered (SLM $\to$ PTS) to minimize computational cost, as the expensive PTS search is performed only on the single best SLM candidate rather than all $U$ candidates.
Joint Spatial-Temporal Phase Optimization (JST-PO): To address the MIMO constraint, a coordinate-descent optimizer jointly adjusts the phase vectors across all $N_t$ transmit antennas and consecutive time slots. The objective function minimizes the worst-antenna PAPR while penalizing rapid changes in phase sequences (differential coding) to reduce side-information entropy.
Relaxed Projection-onto-Convex-Sets (POCS): A final refinement stage applies alternating projections between two convex sets: an amplitude set (clipping peaks above a target $\gamma_{dB}$ ) and a band-limited set (enforcing spectral constraints). A relaxation factor $\delta$ is used to ensure convergence, and the signal is renormalized.

Key Contributions
The paper provides a comprehensive theoretical and simulation-based analysis of this layered approach:

Theoretical Bounds: Derives a PAPR bound for the hybrid waveform and a statistical model for the cascaded SLM-PTS, introducing an "effective number" of independent candidates to account for correlations.
Optimization Proofs: Proves the finite termination of the JST-PO algorithm (a local coordinate-wise minimum) and the convergence of the relaxed POCS iterations to a fixed point.
Distortion Modeling: Applies a Bussgang decomposition to model the residual in-band distortion introduced by the POCS stage, predicting the absence of an error floor compared to traditional clipping.
Side Information Analysis: Quantifies the side-information cost for the cascaded stages and proposes a differential coding scheme for the JST-PO to reduce overhead.
Complexity Quantification: Provides closed-form expressions for computational costs, highlighting that the proposed order (SLM then PTS) is significantly cheaper than a joint search.

Results
Monte Carlo simulations were conducted for a 2 $\times$ 2 Rayleigh MIMO uplink with 32 subcarriers and 8x oversampling. Key findings include:

PAPR Reduction: The proposed scheme reduces the 99.9th-percentile PAPR from approximately 10.2 dB (plain OFDM) to 5.4 dB. This represents a gain of roughly 4.8 dB over OFDM, 2.8 dB over SC-FDMA, 1.5 dB over clipping-and-filtering, and 1.1 dB over $\mu$ -law companding.
Error Performance: In Additive White Gaussian Noise (AWGN), the Bit Error Rate (BER) and Error Vector Magnitude (EVM) of the proposed scheme match those of plain OFDM and SC-FDMA. Crucially, unlike clipping-and-filtering, the proposed scheme does not exhibit an error floor at high SNR.
MIMO Performance: In fading channels, the Block Error Rate (BLER) and BER are marginally better than or comparable to baseline PAPR reduction methods, with no degradation compared to unprocessed OFDM/SC-FDMA.
Efficiency: Under an ideal Class-B PA model, the PAPR reduction translates to a theoretical drain efficiency improvement from ~24% (OFDM) to ~42% (proposed), a factor of 1.74.
Computational Cost: The primary drawback is increased processing time. The proposed scheme requires approximately two orders of magnitude more computation time than clipping or companding baselines in the simulation environment, primarily due to the candidate searches and iterative POCS steps.

Significance and Claims
The authors claim that this work presents a novel, integrated solution that combines a hybrid waveform, a cascaded probabilistic search, a joint multi-antenna optimizer, and a convex projection stage—a combination not previously covered in the cited literature. The significance lies in achieving a "hard" bound on the PAPR tail (via POCS) without incurring the in-band distortion penalties associated with hard clipping. The paper modestly concludes that while the scheme effectively eliminates the error floor and significantly improves PA efficiency potential, the computational overhead currently limits its application to offline or low-rate scenarios, suggesting that future work must focus on low-complexity search strategies and hardware-in-the-loop evaluations.

More like this