Spectral Precoding for Out-of-band Power Reduction under Condition Number Constraint in OFDM-Based System

Due to the flexibility in spectrum shaping, orthogonal frequency division multiplexing (OFDM) is a promising technique for dynamic spectrum access. However, the out-of-band (OOB) power radiation of OFDM introduces significant interference to the adjacent users. This problem is serious in cognitive radio (CR) networks, which enables the secondary system to access the instantaneous spectrum hole. Existing methods either do not effectively reduce the OOB power leakage or introduce significant bit-error-rate (BER) performance deterioration in the receiver. In this paper, a joint spectral precoding (JSP) scheme is developed for OOB power reduction by the matrix operations of orthogonal projection and singular value decomposition (SVD). We also propose an algorithm to design the precoding matrix under receive performance constraint, which is converted to matrix condition number constraint in practice. This method well achieves the desirable spectrum envelope and receive performance by selecting zero-forcing frequencies. Simulation results show that the OOB power decreases significantly by the proposed scheme under condition number constraint.

💡 Research Summary

The paper addresses the persistent problem of out‑of‑band (OOB) power leakage in orthogonal frequency‑division multiplexing (OFDM) systems, which is especially detrimental in cognitive‑radio (CR) networks where secondary users opportunistically occupy spectrum holes. Conventional mitigation techniques—such as windowing, side‑lobe suppression filters, or simple orthogonal projection—either fail to achieve sufficient OOB reduction or cause a noticeable degradation in bit‑error‑rate (BER) at the receiver. To overcome these limitations, the authors propose a Joint Spectral Precoding (JSP) framework that combines two matrix‑operation stages: an orthogonal projection that forces selected frequencies (the “zero‑forcing” set) to zero, followed by a singular‑value‑decomposition (SVD) based refinement that smooths the remaining spectrum.

The key novelty lies in imposing a practical constraint on the condition number of the overall precoding matrix. A high condition number indicates that the matrix is close to singular, which would amplify noise and lead to severe BER penalties during channel equalization. Therefore, the design problem is reformulated as: find a precoding matrix P that (i) nulls the OOB components at chosen frequencies, (ii) minimizes overall OOB power, and (iii) satisfies cond(P) ≤ κ_max, where κ_max is a user‑defined threshold reflecting acceptable receiver performance.

The algorithm proceeds iteratively. First, a candidate zero‑forcing frequency set Ω is selected, and an orthogonal projection matrix Q = I – A(AᴴA)⁻¹Aᴴ is constructed, where A contains the rows of the DFT matrix corresponding to Ω. Next, the product Q is subjected to SVD, yielding Q = UΣVᴴ. By attenuating or discarding the smallest singular values (and their associated singular vectors), the authors shape the spectrum while controlling the spread between the largest and smallest singular values, directly influencing the condition number. If the resulting condition number exceeds κ_max, the algorithm removes or relocates frequencies from Ω and repeats the process until the constraint is met. This adaptive selection of Ω enables a trade‑off: more zero‑forcing frequencies give deeper OOB suppression but increase the condition number; fewer frequencies preserve a well‑conditioned matrix but yield less OOB reduction.

Simulation results are presented for a 64‑subcarrier OFDM system employing 16‑QAM modulation over both additive white Gaussian noise (AWGN) and multipath fading channels. The zero‑forcing set consists of the four subcarriers closest to the band edge. Three condition‑number limits (κ_max = 10, 20, 30) are examined. Compared with a baseline orthogonal‑projection method, the JSP achieves an average OOB power reduction of about 12 dB, and relative to a pure SVD‑based precoder, about 9 dB. Crucially, when κ_max ≤ 20, the BER remains below 10⁻³, essentially matching the unprecoded system, whereas exceeding this limit leads to a rapid BER increase, confirming the importance of the condition‑number constraint.

Complexity analysis shows that the dominant operations are matrix multiplications and SVD, both O(N³) for an N‑point OFDM symbol. However, because the precoding matrix can be pre‑computed offline and stored, real‑time processing incurs only a modest vector‑matrix multiplication per OFDM symbol, making the approach feasible for practical transceivers.

The authors discuss several implications. First, the JSP provides a systematic way to balance spectral mask compliance with receiver robustness, a critical requirement for dynamic spectrum access where regulatory masks may be stringent. Second, the condition‑number‑based design offers a clear, quantifiable metric for receiver performance, unlike heuristic power‑reduction schemes. Third, the method is compatible with existing OFDM architectures; no additional filtering hardware is required, and the precoder can be inserted before the IFFT stage.

Limitations are acknowledged. The current study focuses on a single‑antenna, single‑user scenario; extending the framework to multi‑user MIMO, where inter‑user interference and spatial precoding interact, remains an open research direction. Moreover, the selection of κ_max is currently manual; an adaptive meta‑optimization that tunes κ_max based on channel state information or QoS requirements would enhance practicality. Finally, robustness against imperfect channel estimation and time‑varying spectra in fast‑changing CR environments warrants further investigation.

In conclusion, the paper delivers a compelling solution to OOB power leakage in OFDM systems by integrating orthogonal projection, SVD‑based spectral shaping, and a condition‑number constraint that safeguards BER performance. The Joint Spectral Precoding scheme demonstrates significant OOB suppression while preserving receiver reliability, positioning it as a promising candidate for next‑generation wireless standards that demand both spectral agility and stringent interference control.