采用漫射光束光学的激光二极管可见光通信（LiFi）：系统级建模与交叉验证的ns-3仿真框架
Laser-Diode LiFi With Diffused-Beam Optics: System-Level Modeling and a Cross-Validated ns-3 Simulation Framework
浏览论文内容
中文总结 AI 辅助
本文建立漫射光束LD LiFi收发器系统模型，嵌入ns-3仿真框架，填补原型研究技术差距，仿真显示其在不同距离和调制方式下的传输性能，相关资源已开放复现。
中文摘要 AI 辅助
激光二极管（LD）相比发光二极管在室内光无线接入方面具备一个数量级的带宽优势，但已报道的原型研究常未量化硬件演示与系统级性能之间的差距。本文开发了一套完整、可复现的漫射光束LD LiFi收发器系统模型，该模型包含经全息漫射器整形的500mW激光源、强度调制/直接检测（IM/DD）接收机以及自适应M-QAM信号，并将其嵌入两个经交叉验证的仿真器中：一个开放的ns-3模块，提供全栈网络仿真（涵盖信道、物理层、ARQ MAC、网络设备、IP/UDP/TCP）；另一个用于所有分析误差模型蒙特卡洛验证的Python链路级引擎。从可在14米视距链路传输数据、实时语音及图像的硬件原型出发，本文识别并填补了原型类报告的典型技术差距：将串行接口吞吐量上限误读为光链路容量、缺失噪声建模、无法测量的误差底噪以及未研究的波束宽度/覆盖范围权衡。该框架表明，同一光学前端摆脱2-Mbaud UART瓶颈后，以250-MHz电带宽驱动，在14米距离、3.8×10⁻³ HD-FEC阈值下采用16-QAM可支持930 Mb/s净速率，在5米距离采用256-QAM可扩展至1.86 Gb/s，开关键控调制可支持23.3米传输；20°漫射器在桌面高度覆盖标准房间内半径4.2米的单元。ns-3栈上的网络仿真在70%负载下，饱和有效吞吐量与物理层线路速率差距在7%以内，99百分位延迟低于0.11毫秒。所有模型、代码及图表均已发布以供复现。
英文摘要
Laser diodes (LDs) promise an order-of-magnitude bandwidth advantage over light-emitting diodes for indoor optical wireless access, but reported prototype studies frequently leave the gap between hardware demonstrations and system-level performance unquantified. This paper develops a complete, reproducible system model of a diffused-beam LD LiFi transceiver - a 500-mW laser source beam-shaped by a holographic diffuser, an intensity-modulation/ direct-detection (IM/DD) receiver, and adaptive M-QAM signaling - and embeds it in two cross validated simulators: an open ns-3 module providing full-stack network simulation (channel, PHY, ARQ MAC, Net Device, IP/UDP/TCP) and a Python link-level engine used for Monte Carlo validation of all analytical error models. Starting from a hardware prototype that transferred data, real-time voice, and images over a 14-m line-of-sight link, we identify and close the technical gaps typical of prototype-class reports: serial-interface throughput ceilings misread as optical-link capacity, absent noise modeling, unmeasurable error floors, and unexamined beamwidth/coverage trade-offs. The framework shows that the same optical front end, freed of its 2-Mbaud UART bottleneck and driven at its 250-MHz electrical bandwidth, supports 930 Mb/s net at 14 m under a $3.8 \times 10^{-3}$ HD-FEC threshold with 16-QAM, scales to 1.86 Gb/s at 5 m with 256-QAM, and sustains on-off keying to 23.3 m; a $20^\circ$ diffuser covers a 4.2-m-radius cell of a standard room at desk height. Network simulations over the ns-3 stack yield saturation goodput within 7% of the PHY line rate and sub-0.11-ms 99th-percentile latency at 70% load. All models, code, and figures are released for reproduction.
