End-to-End Autoencoder for Drill String Acoustic Communications

TL;DR

The paper tackles the challenge of achieving low-latency, high-throughput acoustic communications over drill strings. It introduces an end-to-end autoencoder with neural-network-based transmitter and receiver, trained jointly to adapt to the complex drill-string channel. Simulation results show the AE achieving superior BER and PAPR performance at the same throughput, with a ninefold reduction in latency compared to a non-contiguous OFDM baseline. While promising, the approach faces scalability concerns as system size grows, suggesting potential deployment as multiple small AEs multiplexed in frequency for larger bandwidths.

Abstract

Drill string communications are important for drilling efficiency and safety. The design of a low latency drill string communication system with high throughput and reliability remains an open challenge. In this paper a deep learning autoencoder (AE) based end-to-end communication system, where transmitter and receiver implemented as feed forward neural networks, is proposed for acousticdrill string communications. Simulation shows that the AE system is able to outperform a baseline non-contiguous OFDM system in terms of BER and PAPR, operating with lower latency.

Figures (6)

• Figure 1: A frequency response and impulse response obtained using transfer matrix method for a drill string composed of 10 pipes and 9 joints (Table \ref{['tab:drillparams']}).
• Figure 2: Architecture of the autoencoder communication system.
• Figure 3: PSD of signal for the NC-OFDM and the AE trained at $1/N_0 = 3$ dB and $\alpha = 0$.
• Figure 4: BER performance of the baseline NC-OFDM and the AE trained at different $N_0$ and $\alpha=0$.
• Figure 5: PAPR CCDF for the baseline NC-OFDM and the AE trained at $1/ N_0 = 3$ dB and different $\alpha$.