Multi-Label Wireless Interference Classification with Convolutional Neural Networks

Sergej Grunau, Dimitri Block, Uwe Meier

2018 · 30 citations · 6 references

Concepts

TL;DR

The growing use of license‑free frequency bands necessitates reliable coexistence management and proper wireless interference classification (WIC). The study proposes a CNN‑based WIC method that identifies multiple IEEE 802.15.1, IEEE 802.11 b/g, and IEEE 802.15.4 signals amid a utilized signal. The method employs a multi‑label dataset of 10 MHz, 12.8 µs sensing snapshots that pair one utilized signal with up to several interferers. The approach achieves nearly 100 % accuracy on narrow‑band IEEE 802.15.1 and IEEE 802.15.4 interference and over 90 % accuracy on wide‑band IEEE 802.11 b/g cross‑technology interference.

Abstract

The steadily growing use of license-free frequency bands require reliable coexistence management and therefore proper wireless interference classification (WIC). In this work, we propose a WIC approach based upon a deep convolutional neural network (CNN) which classifies multiple IEEE 802.15.1, IEEE 802.11 b/g and IEEE 802.15.4 interfering signals in the presence of a utilized signal. The generated multi-label dataset contains frequency- and time-limited sensing snapshots with the bandwidth of 10MHz and duration of 12.8 μs, respectively. Each snapshot combines one utilized signal with up to multiple interfering signals. The approach shows promising results for same-technology interference with a classification accuracy of approximately 100% for narrow-band IEEE 802.15.1 and IEEE 802.15.4 signals. For cross-technology interference, wide-band IEEE 802.11 b/g signals achieve an accuracy above 90 %.

References

6