Neural Augmentation of Kalman Filter with Hypernetwork for Channel Tracking

TL;DR

The paper introduces Hypernetwork Kalman Filter (HKF), a neural-augmented method that adapts to varying dynamics for improved channel tracking, outperforming traditional Kalman filters especially at high Doppler values.

Contribution

It presents HKF, a novel neural network-based extension of the Kalman filter that adapts to different dynamics without needing multiple filters or prior knowledge.

Findings

HKF matches Kalman filter performance with genie Doppler information.

HKF achieves around 2dB gain over genie Kalman filter at high Doppler.

HKF generalizes well to unseen Doppler, SNR, and pilot patterns.

Abstract

We propose Hypernetwork Kalman Filter (HKF) for tracking applications with multiple different dynamics. The HKF combines generalization power of Kalman filters with expressive power of neural networks. Instead of keeping a bank of Kalman filters and choosing one based on approximating the actual dynamics, HKF adapts itself to each dynamics based on the observed sequence. Through extensive experiments on CDL-B channel model, we show that the HKF can be used for tracking the channel over a wide range of Doppler values, matching Kalman filter performance with genie Doppler information. At high Doppler values, it achieves around 2dB gain over genie Kalman filter. The HKF generalizes well to unseen Doppler, SNR values and pilot patterns unlike LSTM, which suffers from severe performance degradation.