Open Access

ARTICLE

Artificial Neural Network Modeling and LO-CORDIC Multi-Fading Generation for UAV Channel Simulator

Qi Li^1,2, Sathish Kumar Selvaperumal^1,*

1 Faculty of Computing, Engineering & Technology, Asia Pacific University of Technology and Innovation, Kuala Lumpur, Malaysia
2 School of Computer and Communication Engineering, Nanjing Tech University Pujiang College, Nanjing, China

* Corresponding Author: Sathish Kumar Selvaperumal. Email:

(This article belongs to the Special Issue: Aerial Innovation Spectrum: All-Domain Research in UAV Communication, Navigation, and Autonomy)

Computers, Materials & Continua 2026, 89(1), 77 https://doi.org/10.32604/cmc.2026.085073

Received 05 May 2026; Accepted 10 July 2026; Issue published 13 August 2026

Abstract

Unmanned Aerial Vehicle (UAV) air-to-ground (A2G) communication is a core enabling technology for emerging low-altitude wireless applications. At the same time, accurate real-time channel emulation remains a key bottleneck restricting its large-scale engineering deployment. Conventional universal channel simulators exhibit limited fidelity when modeling UAV-specific fading characteristics and degrade real-time performance on resource-constrained hardware platforms. In this study, we develop a dedicated UAV A2G channel simulator based on a heterogeneous FPGA platform (Processing System (PS) + Programmable Logic (PL)). To achieve high-precision path-loss prediction, we train a lightweight backpropagation neural network (BPNN) using field-measured data in agricultural scenarios and deploy the model on the PS for sub-millisecond real-time inference. Experimental results show that the proposed BPNN outperforms the 3GPP TR 38.901, CI, and ray-tracing models, achieving a path-loss prediction RMSE of 1.929 dB. For high-efficiency small-scale fading generation, a locally optimal COordinate Rotation DIgital Computer (LO-CORDIC) algorithm is implemented on the PL, which supports the computation of trigonometric, exponential and logarithmic functions for Rayleigh, Rice and other fading models. The LO-CORDIC reduces the average iterations from 16 to 4.5, achieving a 70.6% reduction in operation latency while maintaining numerical precision. Statistical validation shows that the PDF, mean, and variance of the generated fading signals are highly consistent with theoretical values, with the maximum relative error of 2.12%. This hardware-software co-design architecture effectively balances emulation fidelity, real-time performance and hardware resource consumption. It is suitable for verifying UAV communication algorithms and evaluating hardware-in-the-loop systems.

---

Keywords

---