Wavelength optimisation framework for non-line-of-sight underwater optical wireless communication

Abstract

Underwater optical wireless communication (UOWC) enables high-throughput, low-latency links. However, performance is strongly constrained by wavelength-dependent absorption and scattering in seawater, especially for the non-line-of-sight (NLOS) geometries that arise due to misalignment and platform mobility. This work develops a framework to optimise the wavelength for NLOS UOWC across Jerlov water types – I, III and 9C in the 400–700 nm band. The model employs a Lambertian transmitter, direct and scattered components with Beer–Lambert attenuation, wavelength-dependent silicon photodetector responsivity, and a receiver noise stack (shot and thermal). The electrical SNR, BER (OOK) and channel capacity are investigated. Because the idealised OOK expression can return physically unrealistic raw magnitudes at high SNR, BER is reported against a practical floor of $10^{-12}$ throughout (values below this floor are reported as “$<10^{-12}$” rather than as raw numbers), and results for the most turbid water type (Jerlov 9C) are explicitly flagged as indicative trends rather than quantitative predictions, consistent with the single-scattering approximation underlying the model. Model-based results indicate that the optimum wavelength shifts with turbidity: clear ocean favours the blue–green window (520–570 nm, peak near 564 nm for a silicon photodetector), whereas coastal and turbid waters shift the optimum towards the red band (650–700 nm). This red-shift arises from a combined electro-optical trade-off among absorption, scattering-induced angular spreading, and photodetector responsivity, not from optical attenuation alone. Furthermore, simulation results indicate that distance sensitivity undergoes a gradual change beyond 13 m, rendering NLOS links noise-limited at longer ranges. These model-derived findings offer wavelength-selection and link-budgeting guidance for UOWC system design.