Inter-Cell Interference Rejection Based on Ultrawideband Walsh-Domain Wireless Autoencoding

TL;DR

This work tackles interference from coexisting 5G transmissions in ultrawideband systems and introduces a Walsh-domain end-to-end wireless autoencoder to enhance spectrum sharing. The model jointly optimizes transmitter and receiver mappings in the Walsh domain using 32 parallel branches, with a coding rate $R = \frac{k}{N}$ where $k=4$ and $N=32$, and leverages the self-inverse property of Walsh transforms to distribute information across branches efficiently. Empirical results show the approach can reject up to $12$ dB of inter-cell interference while maintaining BLER near the no-interference baseline for interference near Walsh submultiples of the sampling frequency; performance degrades gracefully as interference moves away from these submultiples. This demonstrates the potential of combining Walsh-domain signal processing with end-to-end learning for dynamic spectrum access in beyond-6G ultrawideband networks, enabling tighter spectrum reuse with legacy 5G systems, and guiding future hardware-aware DL designs for interference-rich regimes. The results leverage $R = rac{k}{N}$ with $k=4$, $N=32$, and show the system can maintain $BLER < 10^{-4}$ under favorable ICI conditions (e.g., near $f_s/2^1$ and $f_s/2^2$).

Abstract

This paper proposes a novel technique for rejecting partial-in-band inter-cell interference (ICI) in ultrawideband communication systems. We present the design of an end-to-end wireless autoencoder architecture that jointly optimizes the transmitter and receiver encoding/decoding in the Walsh domain to mitigate interference from coexisting narrower-band 5G base stations. By exploiting the orthogonality and self-inverse properties of Walsh functions, the system distributes and learns to encode bit-words across parallel Walsh branches. Through analytical modeling and simulation, we characterize how 5G CPOFDM interference maps into the Walsh domain and identify optimal ratios of transmission frequencies and sampling rate where the end-to-end autoencoder achieves the highest rejection. Experimental results show that the proposed autoencoder achieves up to 12 dB of ICI rejection while maintaining a low block error rate (BLER) for the same baseline channel noise, i.e., baseline Signal-to-Noise-Ratio (SNR) without the interference.

Figures (9)

• Figure 1: High-Level Diagram of Research Method Rationale
• Figure 2: Diagram of the transceivers in the End-to-End wireless autoencoder
• Figure 3: Interference pattern in Walsh domain from 2475 MHz downwards
• Figure 4: Interference pattern in Walsh domain downwards from 1250 MHz
• Figure 5: Interference pattern in Walsh domain for different submultiples of the sampling frequency