Out-of-Distribution Detection via Channelwise Feature Aggregation in Neural Network-Based Receivers

TL;DR

This paper introduces a novel layerwise, channelwise feature aggregation method for out-of-distribution detection in neural network-based radio receivers, emphasizing manifold-aware detection aligned with classical receiver behavior.

Contribution

It proposes a post-hoc, layerwise OOD detection framework that avoids classwise statistics and leverages manifold structures, improving detection reliability in multi-label radio receiver outputs.

Findings

Gaussian Mahalanobis with mean activations is the most effective detector.

Earlier layers outperform later layers in OOD detection.

High-delay OOD detection is reliable, high-speed remains challenging.

Abstract

Neural network-based radio receivers are expected to play a key role in future wireless systems, making reliable Out-Of-Distribution (OOD) detection essential. We propose a post-hoc, layerwise OOD framework based on channelwise feature aggregation that avoids classwise statistics--critical for multi-label soft-bit outputs with astronomically many classes. Receiver activations exhibit no discrete clusters but a smooth Signal-to-Noise-Ratio (SNR)-aligned manifold, consistent with classical receiver behavior and motivating manifold-aware OOD detection. We evaluate multiple OOD feature types, distance metrics, and methods across layers. Gaussian Mahalanobis with mean activations is the strongest single detector, earlier layers outperform later, and SNR/classifier fusions offer small, inconsistent AUROC gains. High-delay OOD is detected reliably, while high-speed remains challenging.