Practical Distributed Reception for Wireless Body Area Networks Using Supervised Learning

TL;DR

This paper proposes supervised learning-based distributed reception methods for noncoherent wireless body area networks using on-off keying, enhancing robustness in fast-varying human body channels.

Contribution

It introduces a novel supervised learning approach for symbol detection in noncoherent WBANs, addressing challenges of fast-changing channels without channel state information.

Findings

Supervised learning improves detection accuracy in noncoherent WBANs.

Proposed methods outperform traditional detection techniques in simulations.

Robust communication is achieved despite rapid channel variations.

Abstract

Medical applications have driven many areas of engineering to optimize diagnostic capabilities and convenience. In the near future, wireless body area networks (WBANs) are expected to have widespread impact in medicine. To achieve this impact, however, significant advances in research are needed to cope with the changes of the human body's state, which make coherent communications difficult or even impossible. In this paper, we consider a realistic noncoherent WBAN system model where transmissions and receptions are conducted without any channel state information due to the fast-varying channels of the human body. Using distributed reception, we propose several symbol detection approaches where on-off keying (OOK) modulation is exploited, among which a supervised-learning-based approach is developed to overcome the noncoherent system issue. Through simulation results, we compare and verify the performance of the proposed techniques for noncoherent WBANs with OOK transmissions. We show that the well-defined detection techniques with a supervised-learning-based approach enable robust communications for noncoherent WBAN systems.