MI-ISAC: Magneto-Inductive Integrated Sensing and Communication in the Reactive Near-Field for RF-Denied Environments

TL;DR

MI-ISAC tackles sensing and communication in RF-denied environments by exploiting reactive near-field magneto-inductive coupling. The approach relies on a deterministic, geometry-driven MI channel with a $h(r,\theta,\phi)=\frac{C}{r^{3}} g(\theta,\phi)$ form and a 3×3 MI-MIMO system built from tri-axial coils, enabling gradient-based ranging that can reach sub-millimeter accuracy at typical MI ranges. Five foundational insights are established: identifiability requires tri-axial coils (rank 3, $\kappa=2$), the CRB scales as $r^{8}$, ToF is impractical at MI bandwidths, MI-ISAC yields 4–10+ dB sensing gain over TDMA, and the MI-MIMO channel has a universal $\{+2,-1,-1\}$ eigenstructure. The framework supports applications in underground, underwater, and in-body domains, and outlines a roadmap for physics, processing, and networking challenges toward practical deployment. Overall, MI-ISAC provides a pathway to integrated sensing and communication in environments where RF propagation is severely attenuated, enabling 6G-ready capabilities with minimal sensing overhead.

Abstract

Radio-frequency integrated sensing and communication (RF-ISAC) is ineffective inunderground, underwater, and in-body environments where conductive media attenuate electromagnetic waves by tens of dB per meter. This article presents magneto-inductive ISAC (MI-ISAC), a paradigm that exploits the reactive near-field quasi-static coupling inherent to MI links, enabling a fundamentally different approach to ISAC in these RF-denied environments. Five foundational results are established: (i)~tri-axial coils are necessary and sufficient for identifiable joint range-and-angle estimation; (ii)~coupling strength changes sharply with range, enabling theoretical sub-millimeter accuracy at typical MI distances despite kHz-level bandwidth; (iii)~time-of-flight is ineffective under such narrow bandwidth, but the coupling gradient provides approximately six orders of magnitude finer resolution; (iv)~MI-ISAC can provide 4--10+\,dB sensing gain over time-division baselines; and (v)~the MI-MIMO channel is geometry-invariant and well-conditioned across all orientations. Applications and a research roadmap are discussed.

Figures (5)

• Figure 1: MI-ISAC concept: three RF-denied environments (underground, underwater, in-body) unified by a single reactive near-field coupling link. The central MI-ISAC loop illustrates how communication, channel-based sensing, parameter estimation, and closed-loop control share the same waveform and hardware through the geometry-deterministic $3\!\times\!3$ MI-MIMO channel.
• Figure 2: MI channel model and tri-axial coil geometry. Two nodes with local coordinate frames $(X_t,Y_t,Z_t)$ and $(X_r,Y_r,Z_r)$ are separated by vector $\bm{r}$. The coupling tensor $\bm{G} = 3\hat{\bm{r}}\hat{\bm{r}}^{\!\top}-\bm{I}_3$ has eigenvalues $\{+2,-1,-1\}$ and corresponding radial/tangential eigenmodes.
• Figure 3: (a) FIM rank: single-axis (rank = 1, not identifiable) vs. tri-axial (rank = 3, identifiable). (b)$\sqrt{\mathrm{CRB}(r)}$ vs. distance showing $r^{4}$ scaling (equivalently $r^{8}$ in CRB). Blue: ideal thermal noise; red: practical front-end (NF = 6 dB, finite-$Q$ loss = 3 dB). MC MLE markers validate both bounds. At $r\!=\!10$ m: 0.1 mm (ideal) vs. 0.3 mm (practical)---both sub-millimeter. Parameters: $a\!=\!0.15$ m, $N_t\!=\!20$, $f_0\!=\!10$ kHz, $N\!=\!100$, $P\!=\!1$ W, $B\!=\!1$ kHz.
• Figure 4: Resolution comparison: MI coupling-based (RSSI) vs. RF ToF at various bandwidths. Within the MI communication range ($r < r^{\star}$), MI achieves over three orders of magnitude finer resolution than 500 MHz UWB. The paradigm shift: from bandwidth-limited ToF to gradient-limited RSSI.
• Figure 5: MI-ISAC research roadmap: a $3\!\times\!3$ taxonomy spanning Physics & Hardware, Signal Processing & ISAC Design, and Networking & Applications across near-term (1--2 yr), mid-term (3--5 yr), and long-term (5+ yr) horizons. Mode B (self-impedance sensing) is a future extension.