Pulse-Shaped OFDM (PS-OFDM)

• Pulse-Shaped OFDM (PS-OFDM) is a multicarrier waveform that replaces the traditional rectangular pulse with a tailored shaping pulse to improve spectral localization and reduce out-of-band emissions.
• It employs methods such as windowing, frequency-domain filtering, and DFT spreading to achieve optimal time-frequency localization, minimize ISI/ICI, and lower PAPR.
• Recent implementations demonstrate up to 60–70 dB out-of-band suppression and a 1–3 dB reduction in PAPR, making PS-OFDM critical for advanced wireless and powerline communication standards.

Pulse-Shaped OFDM (PS-OFDM) refers to a broad class of multicarrier waveforms in which the canonical rectangular subcarrier pulse of OFDM is replaced with a more general shaping pulse, improving spectral localization, reducing out-of-band emissions (OOBE), and often enabling enhanced robustness and lower peak-to-average power ratio (PAPR). PS-OFDM encompasses methods ranging from simple windowing to fully parametric frequency- or time-domain filter designs, and has become foundational in advanced wireless and powerline communication standards where spectral efficiency, regulatory mask compliance, PAPR, and flexibility are critical.

1. Signal Models and Pulse-Shaping Fundamentals

The core construction of PS-OFDM is the synthesis of the transmit signal as a weighted sum of shifted, shaped subcarrier pulses. For a generic baseband system, the signal can be written:

$$x(t) = \sum_{n=-\infty}^{\infty} \sum_{k=0}^{M_A-1} a_{n,k}\;g(t-nT)\,e^{j2\pi kF(t-nT)}$$

where $g(t)$ is the prototype pulse, $T$ the OFDM symbol duration (with or without a guard interval), $F$ the subcarrier spacing, and $a_{n,k}$ the data symbols. With rectangular $g(t)$, classical CP-OFDM is recovered; with more general $g(t)$, various PS-OFDM designs arise (Zhao et al., 2016).

PS-OFDM can equivalently be realized by applying windowing or filtering in the time or frequency domain to standard OFDM blocks. For example, in powerline systems, PS-OFDM is implemented as a time-domain window (typically raised-cosine or root-raised-cosine) overlapping adjacent OFDM symbols (Girotto et al., 2016).

In advanced variants, shaping can also be imposed via frequency-domain filtering after subcarrier mapping or via subband-level DFT precoding, as in DFT-s-OFDM and circularly pulse-shaped OFDM (CPS-OFDM) (Huang et al., 2018).

2. Pulse Design Objectives and Parameterization

Pulse design in PS-OFDM balances spectral containment, orthogonality, time-frequency localization, and implementation feasibility. Crucial criteria include:

• Zero-ISI/ICI Conditions: Perfect reconstruction (zero inter-symbol and inter-carrier interference) via time-frequency orthogonality or bi-orthogonality, formalized as $$\langle g_{m',n'},\gamma_{m,n}\rangle=\delta_{m'm}\,\delta_{n'n}$$ for properly chosen transmit ($g$) and receive ($\gamma$) pulses (Üçüncü et al., 2015).
• Spectral Confinement: Suppression of OOBE to meet regulatory masks or minimize adjacent channel interference. Achieved by optimizing $g(t)$, window length, or combined time- and frequency-domain shaping terms (Girotto et al., 2016, Díez et al., 2018).
• Time-Frequency Localization (TFL): Minimized Heisenberg uncertainty product $\xi$ for ISI/ICI robustness.
• Envelope Fluctuation Reduction: Lower PAPR or related metrics such as cubic metric (CM) or variance of instantaneous power (VIP), to enable efficient power amplification (Carpi et al., 2024, Huang et al., 2018, Huang et al., 2018).

Pulse parameterization methods include analytic forms (Gaussian, RRC, RC), orthogonalization via Zak transform, numerical solutions of convex/quadratic programs, and learned polynomial models in the frequency domain (Carpi et al., 2024, Üçüncü et al., 2015, Zhao et al., 2016).

3. Major PS-OFDM Design Frameworks and Optimization Methods

3.1 Windowing and Overlap-Add

Windowed overlap-add approaches (e.g., raised-cosine, vestigial symmetry, asymmetric windows) suppress OOBE by smoothing the transitions at OFDM symbol edges (Nguyen et al., 2017). The roll-off duration is typically limited to the CP length to prevent ISI, and asymmetric or vestigial symmetry windowing can further improve spectral performance without ISI penalties (Nguyen et al., 2017).

3.2 Time and Frequency-Domain Spectral Shaping

Generalized pulses are constructed as linear combinations of the standard pulses, active interference cancellation (AIC) carriers, and adaptive symbol transition (AST) pulses. Data-independent quadratic programs allow for offline design, yielding analytically closed-form PSDs and transparent integration in OFDM transceivers (Díez et al., 2018, Giménez et al., 30 Dec 2025). With the use of Hermitian-symmetric or half-complex pulse designs, the number of real optimization variables and implementation multiplications is halved, achieving identical spectral performance at markedly reduced computational cost (Giménez et al., 5 Nov 2025).

3.3 Circular/DFT-spread and Learned Shaping

Circular pulse shaping (CPS-OFDM) and DFT-s-OFDM with frequency-domain spectrum shaping (FDSS) introduce subband DFT-based precoding structures or parametric frequency-domain masks. These architectures enable joint minimization of PAPR/CM, OOBE, and error metrics under energy, spectral flatness, and zero-ISI constraints. Learning-based design (e.g., gradient-based optimization of frequency-domain polynomials) further adapts to changing system or mask parameters and facilitates flexible resampling or retraining for different bandwidth allocations (Carpi et al., 2024, Huang et al., 2018, Huang et al., 2018).

3.4 SINR-Optimal "Ping-Pong" Optimization

POPS-OFDM (Ping-Pong Optimized Pulse Shaping OFDM) alternates the design of transmit and receive pulses by maximizing SINR over actual time-frequency dispersive channel models. Each step solves a generalized Rayleigh quotient eigenproblem, converging to bi-orthogonal pulse pairs with high SIR and low OOB emissions across realistic channel conditions (Hraiech et al., 2015).

4. Regulatory and System-Level Constraints

PS-OFDM finds strong justification in environments with stringent spectrum emission requirements (e.g., PLC, TV white space, cognitive radio). Detailed analyses of EMC masks (e.g., EN 50065, EN 50561-1) and power spectral density measurement procedures demonstrate PS-OFDM's efficacy in broadband/narrowband PLC and its centrality in IEEE P1901/1901.2 standards (Girotto et al., 2016).

PS-OFDM methods such as active interference cancellation, adaptive symbol transition, multi-carrier spectral notching, and joint time-frequency shaping are pivotal to achieving deep spectral notches (≥40–60 dB OOB suppression), minimal data carrier loss, and full compliance under noncontiguous and dynamically variable mask profiles (Díez et al., 2018, Giménez et al., 30 Dec 2025).

5. Complexity, Implementation, and Practical Aspects

PS-OFDM adds marginal computational complexity over CP-OFDM—typically 1–2% for short polyphase network filters or window functions, scaling to ~16% when full generalized pulses or large AST dictionaries are mapped (Zhao et al., 2016, Díez et al., 2018). Blockwise operations, real-only optimization (Hermitian symmetry), and precomputed offline pulse dictionaries contribute to low run-time cost (Giménez et al., 5 Nov 2025, Giménez et al., 30 Dec 2025).

The receiver for most PS-OFDM schemes remains backward-compatible, often requiring no changes except for possible equalizer updates to absorb circular shifts or constant phase rotations. The method is fully compatible with FFT-based or polyphase hardware architectures, amenable to FPGA/ASIC realization (Giménez et al., 5 Nov 2025, Zhao et al., 2016).

6. System Performance, Trade-Offs, and Evaluation

6.1 Spectral/Efficiency/Flexibility

PS-OFDM achieves steeper spectral roll-off (e.g., sidelobe decay of $1/f^6$ or exponential with RRC or Gaussian pulses) and can attain regulatory mask compliance with minimal or no guard band or null subcarrier overhead (down from 27–28% guard to sub-4% using joint AIC+AST) (Díez et al., 2018, Girotto et al., 2016). Tables from (Huang et al., 2018) and (Huang et al., 2018) document 60–70 dB subband OOB suppression, with PAPR reduced by 1–3 dB over competing DFT-S-OFDM or windowed OFDM designs.

6.2 Robustness and Multi-Service Support

Time-frequency localization improves resilience to delay spread and Doppler; e.g., PS-OFDM tolerates ±15% timing offset, ±2% Doppler, compared to CP-OFDM's ±7%/±1% limits (Zhao et al., 2016). This renders PS-OFDM particularly well-suited to asynchronous multiuser access, mixed numerology, and highly mobile environments (Huang et al., 2018).

6.3 PAPR, CM, and Nonlinear Distortion

PS-OFDM methods suppress PAPR (and, for CPS-OFDM, cubic metric) by 2–4 dB (cf. rectangular) with only minor (≤0.05 dB) SNR penalty at typical $BER=10^{-2}$–$10^{-3}$ operation points (Carpi et al., 2024, Kamruzzaman et al., 2010). Optimization under EVM, OSBE, and SE constraints further supports high PA efficiency and increased spectral efficiency in highly regulated or nonlinear settings (Huang et al., 2018).

6.4 Application Highlights

• NB/BB-PLC: PS-OFDM median capacity ≈30 kbps (NB), ≈270 Mbps (BB), achievable via standard-compliant RC windowing (Girotto et al., 2016).
• IEEE 802.11af: Asymmetric pulse shaping enables >60 dB sidelobe suppression with ≈0.1 ISI/SE penalty over classical windows (Nguyen et al., 2017).
• 5G NR: CPS-OFDM outperforms both OFDMA and DFT-s-OFDM on OOBE, PAPR, and robustness; able to deliver ∼4.8 bits/s/Hz spectral efficiency with 60+ dB OOB suppression (Huang et al., 2018).

7. Open Challenges and Future Directions

Open technical issues include the derivation of analytic ISI bounds under moderate PAPR constraints, learning-based pulse design under real-time, time-varying channel statistics, integration with hierarchical (3GPP) codebooks, and streaming implementations that support sub-millisecond reconfiguration for cognitive radio or dynamic spectrum environments (Carpi et al., 2024, Giménez et al., 30 Dec 2025). Fully co-optimized time/frequency and nonlinear-domain pulse-shaping (e.g., for PA nonlinearity-aware design) remains an area of ongoing interest (Huang et al., 2018).

A plausible implication is that PS-OFDM, via its highly general and modular framework, is expected to remain a central tool in future 5G-and-beyond physical layer standards, especially where spectrum flexibility, emission regulations, and hardware efficiency intersect or where multi-service coexistence is paramount.