Visible light communication is being explored as one way to expand the capacity of future wireless networks by turning light sources into data transmitters. But the physical layer that makes optical communication practical has its own computational costs. A new study published on 4 October 2026 proposes a different form of optical orthogonal frequency-division multiplexing that substantially reduces those processing demands while retaining comparable error performance in simulations.
The researchers call the method Binary-Sign Optical OFDM, or BSO-OFDM. At a fixed spectral efficiency of 2.5 bits per second per hertz, their theoretical and simulation results indicate that the approach reduced computational complexity by more than 75% compared with asymmetrically clipped optical OFDM, or ACO-OFDM, and by more than 45% compared with direct-current-biased optical OFDM, or DCO-OFDM, and Flip-OFDM.
The work matters because the push toward sixth-generation, or 6G, networks is not only about increasing transmission rates. Networks designed to support artificial intelligence at many points in the system may also face growing processing demands at access points, devices and edge nodes. A physical layer that requires fewer mathematical operations could therefore reduce one part of that computational burden, particularly in optical wireless systems.
Why optical OFDM carries a processing cost
Conventional radio-frequency OFDM can represent information using complex-valued signals. Intensity-modulation and direct-detection visible light systems face a different constraint: the transmitted waveform must ultimately be real and non-negative because information is carried through variations in optical intensity.
Existing optical OFDM methods solve this problem in different ways, but those solutions can consume signal resources or add processing stages. Hermitian symmetry can be imposed so that an inverse Fourier transform produces a real-valued waveform. Other methods add a direct-current bias, clip portions of the signal, split positive and negative components across transmissions, or combine several of these techniques.
According to the researchers, these operations create a particularly relevant design problem for AI-native 6G systems. If increasingly intelligent networks demand more processing elsewhere, reducing avoidable physical-layer computation could help keep the overall architecture manageable.
A different way to represent the optical signal
BSO-OFDM approaches the real and non-negative transmission requirement differently. Rather than imposing Hermitian symmetry, it separates information about the absolute value of the signal from binary information describing its sign. This lets the system use the available subcarriers without relying on the conventional symmetry construction.
The proposed architecture also avoids the DC bias used in DCO-OFDM and the asymmetric clipping associated with ACO-OFDM. These design choices are central to the claimed reduction in computational complexity rather than being secondary optimisations added after the waveform is created.
Mohammed K. Mohaisen, Mohd Fadzli Mohd Salleh and Samir M. Hameed evaluated the system theoretically and through simulations. Their comparisons held spectral efficiency at 2.5 bits/s/Hz so that differences in processing and transmission behaviour were not simply the result of allowing one scheme to carry less information per unit bandwidth.
Complexity fell sharply against three established alternatives
The largest reported advantage concerned computational complexity. BSO-OFDM required more than 75% less computation than ACO-OFDM in the authors’ comparison. Relative to DCO-OFDM and Flip-OFDM, the reduction was more than 45%.
These are substantial differences because fast Fourier transforms, inverse transforms and the additional processing needed to make optical OFDM signals physically transmissible are repeated continuously as data move through a communication system. A reduction at the waveform-processing level can therefore matter even when individual operations are already fast on modern hardware.
The authors also examined peak-to-average power ratio, or PAPR. High PAPR is a longstanding challenge in OFDM because occasional large signal peaks can push transmitters toward nonlinear operating regions or require greater dynamic range. The proposed scheme achieved roughly a 4 dB PAPR reduction compared with ACO-OFDM and Flip-OFDM.
Lower complexity did not require a large BER sacrifice
A simpler waveform would be much less useful if it substantially increased transmission errors. The study therefore evaluated bit error rate, or BER, across several channel conditions.
The researchers tested BSO-OFDM over an additive white Gaussian noise channel, a multipath channel with additive white Gaussian noise, and a physical visible-light channel incorporating line-of-sight and diffuse propagation. Across these evaluations, the proposed system maintained BER performance similar to the comparison methods while showing improved robustness under multipath conditions.
That combination is important to the paper’s argument. The result is not merely that fewer calculations can be performed. Rather, the simulations suggest that the authors can remove or reorganise substantial parts of the conventional optical OFDM processing chain without giving up the error-rate performance that makes OFDM attractive in the first place.
What the results could mean for visible light communication
Visible light communication has an unusual advantage: lighting infrastructure can potentially serve a second function as a communication medium. Optical wireless links could complement radio networks in indoor environments and other settings where dense connectivity, electromagnetic compatibility or additional spectrum are valuable.
The new study addresses a less visible part of that proposition. A future optical network still needs practical signal processing, and the cost of that processing matters if communication is to be integrated into increasingly complex intelligent systems.
The reported reductions do not mean that BSO-OFDM will necessarily reduce the total energy consumption of a deployed 6G network by the same percentages. Computational-complexity counts describe the algorithmic workload of the compared physical-layer schemes, not the complete power draw of processors, optical transmitters, receivers, lighting systems and networking equipment. Hardware implementation can also change the relative cost of different mathematical operations.
The evidence is promising but remains pre-deployment
The main limitation is that the reported performance is based on theoretical analysis and simulation rather than a large-scale deployed 6G visible-light network. Real optical channels introduce hardware nonlinearities, imperfect synchronisation, component limitations, changing illumination conditions, interference and implementation overhead that models can represent only to varying degrees.
The fixed spectral efficiency of 2.5 bits/s/Hz also makes the comparison controlled, but future work will need to establish how the scheme behaves across a wider range of data rates, modulation orders, optical environments and hardware configurations. The phrase AI-native 6G describes the intended future context rather than an existing commercial network in which BSO-OFDM has already been demonstrated.
Even with those qualifications, the study presents a clear engineering trade-off worth testing experimentally. BSO-OFDM reduced the calculated processing burden sharply, lowered PAPR against two important alternatives and maintained broadly similar BER performance across the modelled channels. If those advantages survive implementation in physical hardware, the method could offer a leaner optical physical layer for future high-density wireless systems.
Source Information
Study: A low-complexity optical OFDM physical layer for AI-native 6G visible light communication networks
Authors: Mohammed K. Mohaisen, Mohd Fadzli Mohd Salleh and Samir M. Hameed
Journal: Scientific Reports
Published: 4 October 2026
DOI: 10.1038/s41598-026-74508-1
Study type: Theoretical and simulation-based communications engineering study
