WiRM: Dual Applications in Astrophysics & Wireless Sensing

• WiRM is a dual-use term that, in astrophysics, links Faraday rotation and emission measures to infer magnetic fields in the warm ionized medium, and in wireless sensing, it denotes a contactless respiration monitoring framework using Wi-Fi CSI.
• In the astrophysical context, WiRM leverages empirical relations from MHD simulations to connect RM dispersion and EM statistics with the sonic Mach number and line-of-sight magnetic field strength.
• In wireless sensing, WiRM employs conjugate CSI processing and fast iterative filtering to accurately extract respiratory rate and waveform from noisy Wi-Fi signals.

Searching arXiv for papers on "WiRM" to ground the article and disambiguate the term. WiRM denotes two distinct research usages. In Galactic interstellar-medium studies, the term is a natural contraction of “WIM + RM,” referring to the use of Faraday rotation measures to infer magnetic-field properties in the warm ionized medium; in this sense, it functions as a compact label for a statistical inference toolkit linking observables such as RM and EM to turbulent parameters and the line-of-sight regular magnetic field (Wu et al., 2014). In wireless sensing, WiRM is the title of “WiRM: Wireless Respiration Monitoring Using Conjugate Multiple Channel State Information and Fast Iterative Filtering in Wi-Fi Systems,” a two-staged framework for contactless respiration monitoring from commodity Wi-Fi CSI that estimates both respiratory rate and respiratory waveform (Rhodes et al., 31 Jul 2025). The coexistence of these usages is terminological rather than conceptual: one belongs to Galactic magneto-ionic turbulence, the other to RF sensing and biomedical signal processing.

1. WiRM as a term: dual usage and disciplinary separation

In astrophysical usage, “WiRM” naturally evokes “WIM + RM,” namely the use of Faraday rotation measures to study the magnetic field in the warm ionized medium. The associated methodology is centered on empirical relations connecting the turbulent properties of the WIM, especially the sonic Mach number, and magnetic-field strength to RM and EM statistics (Wu et al., 2014). This usage is rooted in the broader modern view of the WIM as a major, physically distinct component of the Galactic interstellar medium, extending more than $1\,\mathrm{kpc}$ from the midplane and characterized by diffuse ionized gas with temperatures of order $10^4$ K, a large filling factor, and diagnostic optical line ratios that differ systematically from classical H II regions (Hill et al., 2012).

In wireless sensing, WiRM is an acronym for Wireless Respiration Monitoring. Here the term refers to a specific algorithmic framework that uses channel state information in Wi-Fi systems to perform contactless respiration monitoring, with the explicit goal of estimating both the instantaneous respiratory rate and the respiratory waveform (Rhodes et al., 31 Jul 2025).

The two usages share only a surface lexical overlap. A plausible implication is that literature searches for “WiRM” require immediate contextual disambiguation, since the underlying technical objects—Faraday rotation statistics in turbulent plasma and CSI-based respiratory sensing in OFDM systems—are unrelated.

2. Astrophysical WiRM: warm ionized medium and Faraday rotation

The astrophysical usage is anchored in the warm ionized medium, defined as the diffuse, warm, ionized component of the Galactic ISM. The WIM is described as an extended, low-density plasma of almost fully ionized hydrogen pervading the Galactic disk and halo, with temperatures of order $10^4$ K, electron densities typically $\ll 1\ \mathrm{cm}^{-3}$, and a vertical extent exceeding $1$ kpc from the Galactic plane (Hill et al., 2012). One summary gives typical values $\bar{T} \sim 8000$ K and $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$, with a plane-parallel layer of scale height $\sim 2$ kpc and filling $\sim 20\text{–}30$% of the volume in that layer (Wu et al., 2014).

Its observational distinctness is emphasized by pulsar dispersion measures, free–free absorption, diffuse H$\alpha$, and characteristic optical line ratios. The WIM is physically distinct from classical H II regions because of systematically different temperatures, ionization states, and line ratios, including enhanced $10^4$0 and $10^4$1, weaker $10^4$2, and weaker $10^4$3 (Hill et al., 2012). The [N II] $10^4$4 temperature diagnostic yields $10^4$5 K for the WIM and $10^4$6 K for classical H II regions, as summarized in the review (Hill et al., 2012).

Within this environment, RM and EM serve as complementary probes. The standard definitions are

$10^4$7

$10^4$8

and

$10^4$9

The electron-density-weighted LOS field is

$10^4$0

RM is therefore sensitive to $10^4$1 and the line-of-sight magnetic field $10^4$2, whereas EM traces $10^4$3 and hence density fluctuations and compressibility (Wu et al., 2014).

This is the core of the astrophysical WiRM idea: use EM statistics to estimate turbulent compressibility, then use RM statistics conditioned on that compressibility to estimate the LOS regular magnetic field (Wu et al., 2014).

3. Empirical WiRM relations in turbulent WIM studies

The principal formulation appears in “Magnetic Field and Faraday Rotation Measure in the Turbulent Warm Ionized Medium” (Wu et al., 2014). The study extends earlier work by incorporating Mach-number dependence into the relation between LOS magnetic field strength and the dispersion of RM distribution in turbulent media with root-mean-square sonic Mach number $10^4$4. The simulations cover $10^4$5, the Mach-number range identified as appropriate for the Galactic WIM (Wu et al., 2014).

The analysis uses the full width at half maximum of the frequency distribution of the normalized RM, denoted

$10^4$6

where $10^4$7 is the mean RM. Since the distribution is Gaussian, the relation between FWHM and standard deviation is

$10^4$8

For EM, the study examines the distribution of $10^4$9, which is Gaussian, so EM itself is well described by a log-normal distribution; its width is quantified by

$\ll 1\ \mathrm{cm}^{-3}$0

The main empirical relation, their Equation 12, is

$\ll 1\ \mathrm{cm}^{-3}$1

for the fiducial WIM parameters $\ll 1\ \mathrm{cm}^{-3}$2 K and $\ll 1\ \mathrm{cm}^{-3}$3 (Wu et al., 2014). For other values, the field scales as

$\ll 1\ \mathrm{cm}^{-3}$4

A second empirical relation, their Equation 13, links sonic Mach number to EM dispersion: $\ll 1\ \mathrm{cm}^{-3}$5 The paper states that this is grounded in the known relation between density dispersion and Mach number in isothermal turbulence,

$\ll 1\ \mathrm{cm}^{-3}$6

with $\ll 1\ \mathrm{cm}^{-3}$7 for solenoidal forcing (Wu et al., 2014).

Together these relations constitute the astrophysical WiRM workflow: estimate $\ll 1\ \mathrm{cm}^{-3}$8 from EM statistics, then infer $\ll 1\ \mathrm{cm}^{-3}$9 from RM dispersion and the inferred Mach number. The paper explicitly characterizes these as relations that “could be used for a quick and rough estimation of the LOS magnetic field strength in the turbulent WIM” (Wu et al., 2014).

4. Numerical basis, physical interpretation, and limitations of the WIM+RM framework

The empirical relations are derived from three-dimensional, magnetohydrodynamic isothermal turbulence simulations with solenoidal forcing (Wu et al., 2014). The numerical setup consists of compressible isothermal MHD solved with a TVD scheme, a periodic cubic box with $1$0 grid cells, and purely solenoidal forcing generated in Fourier space with $1$1. Equal-power forcing is applied in the narrow wave-number range $1$2, with amplitude chosen to maintain constant kinetic energy input rate (Wu et al., 2014).

The parameter space includes $1$3 and $1$4, giving 9 simulations total. Turbulence saturates around $1$5, where $1$6, and the simulations run to $1$7, with 11 snapshots taken in the saturated stage (Wu et al., 2014).

For WIM fiducial values $1$8 K and $1$9, the initial magnetic field strength is

$\bar{T} \sim 8000$0

giving $\bar{T} \sim 8000$1 for $\bar{T} \sim 8000$2, $\bar{T} \sim 8000$3 for $\bar{T} \sim 8000$4, and $\bar{T} \sim 8000$5 for $\bar{T} \sim 8000$6 (Wu et al., 2014).

The physical interpretation is presented in terms of Alfvénic perturbations. In a weak $\bar{T} \sim 8000$7–$\bar{T} \sim 8000$8 correlation regime,

$\bar{T} \sim 8000$9

and the paper argues that

$\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$0

so that

$\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$1

The empirical fit modifies the exponents to $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$2 and $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$3 rather than exactly unity to match the simulation data (Wu et al., 2014).

The authors also emphasize limitations. The relations are derived for isothermal MHD turbulence, only solenoidal forcing is considered, the Mach-number range is limited to $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$4, and the simulations are homogeneous boxes without Galactic stratification or multiphase structure (Wu et al., 2014). They further note that the method requires $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$5 not to be too small, since the relation breaks down as $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$6 and $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$7 (Wu et al., 2014). The review of the WIM reinforces that the real medium is a galaxy-wide, extended ionized layer with a substantial, but subdominant, scattered-light contribution to faint high-latitude H$\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$8, rather than a simple homogeneous slab (Hill et al., 2012). This suggests that astrophysical WiRM is best regarded as a statistical inference tool for WIM-dominated regions rather than a precision diagnostic for arbitrary individual sightlines.

5. WiRM in wireless sensing: CSI-based respiration monitoring

The wireless-sensing usage is defined by “WiRM: Wireless Respiration Monitoring Using Conjugate Multiple Channel State Information and Fast Iterative Filtering in Wi-Fi Systems” (Rhodes et al., 31 Jul 2025). The framework is described as a two-staged approach to contactless respiration monitoring. In the first stage it improves respiratory rate estimation by using conjugate multiplication for phase sanitisation and adaptive multi-trace carving for tracing how the respiratory rate changes over time; in the second stage it uses the improved respiratory rate estimate to inform the decomposition and selection of the respiratory waveform from the CSI data (Rhodes et al., 31 Jul 2025).

The problem formulation begins with noisy CSI measurements affected by phase noise, thermal noise, multiplicative channel noise, and multipath. For one transmit antenna $\bar{n}_e \sim 0.03\ \mathrm{cm^{-3}}$9, one receive antenna $\sim 2$0, subcarrier $\sim 2$1, and time $\sim 2$2, the ideal channel with one breathing-affected path is modeled as

$\sim 2$3

and the noisy CSI as

$\sim 2$4

The task is to recover both the time-varying respiratory frequency and the underlying respiratory waveform $\sim 2$5 (Rhodes et al., 31 Jul 2025).

Stage 1 forms conjugate multiple CSI links,

$\sim 2$6

in order to cancel common phase noise from antennas sharing an oscillator (Rhodes et al., 31 Jul 2025). Linear ACFs of magnitude and phase are then computed, lag $\sim 2$7 is discarded, the resulting traces are weighted by breathing-to-noise ratio, and the weighted ACFs are combined into a single signal whose zoom FFT over the respiratory band yields a spectrogram. Adaptive Multi-Trace Carving is then applied to obtain a smooth time-varying respiratory-rate trace and a breath-presence indicator (Rhodes et al., 31 Jul 2025).

Stage 2 collects a longer window, performs conjugate multiplication again, extracts magnitude and phase, applies z-score normalization, and stacks the normalized traces into a large matrix. The latest rate estimate from Stage 1 is then used as a prior for subcarrier selection through

$\sim 2$8

with the primary subcarrier chosen by

$\sim 2$9

After phase-aligned, magnitude-weighted combining, the resulting signal is decomposed by Fast Iterative Filtering into intrinsic mode functions,

$\sim 20\text{–}30$0

and the breathing IMF is selected as the component whose dominant frequency is closest to the Stage 1 rate estimate (Rhodes et al., 31 Jul 2025).

This architecture explicitly couples rate estimation and waveform reconstruction. The paper states that most prior Wi-Fi methods either estimate only rate or attempt waveform extraction with fragile decomposition heuristics; WiRM instead uses the improved rate estimate to guide waveform extraction (Rhodes et al., 31 Jul 2025).

6. Evaluation, robustness, and the relation between the two WiRM usages

The wireless WiRM framework is evaluated on the rf_polysomnography_respiration_data_00 public dataset with commodity Wi-Fi CSI, OFDM, 114 subcarriers, sampling frequency $\sim 20\text{–}30$1, 2 transmit antennas, and 2 receive antennas, using clinical polysomnography as ground truth (Rhodes et al., 31 Jul 2025). Respiratory rate is evaluated by RMSE in BPM and by the percentage of time samples with error below 3 BPM, while waveform quality is evaluated by sliding-window absolute correlation with the ground-truth respiratory waveform (Rhodes et al., 31 Jul 2025).

Compared against three state-of-the-art methods, WiRM achieves an average reduction of $\sim 20\text{–}30$2 in respiratory rate root mean squared error and an average improvement of $\sim 20\text{–}30$3 in the fraction of estimates within 3 BPM (Rhodes et al., 31 Jul 2025). For waveform extraction, the paper reports a $\sim 20\text{–}30$4 improvement in average absolute correlation with the ground truth respiratory waveform and a $\sim 20\text{–}30$5 improvement in maximum absolute correlation, relative to Pos-Free Breath (Rhodes et al., 31 Jul 2025). On one illustrated night, WiRM correctly detects breath presence in $\sim 20\text{–}30$6 of time (Rhodes et al., 31 Jul 2025).

The paper also introduces a purpose-built simulation toolkit to evaluate robustness under thermal, multiplicative, and phase noise. In the phase-noise simulations, WiRM achieves 0.0 BPM rate RMSE and waveform correlation $\sim 20\text{–}30$7; in multiplicative-noise simulations, it maintains 0.0 BPM RMSE and $\sim 20\text{–}30$8 across the tested standard deviations 0.1–1.0; under high thermal noise, it degrades but remains more resilient than several baselines (Rhodes et al., 31 Jul 2025). The authors therefore conclude that WiRM demonstrates improved or comparable resilience to these common noise sources (Rhodes et al., 31 Jul 2025).

The astrophysical and wireless senses of WiRM differ in object, methodology, and validation standard. The first links RM and EM statistics to $\sim 20\text{–}30$9 and $\alpha$0 through MHD simulation-derived empirical relations in the Galactic WIM (Wu et al., 2014); the second links CSI preprocessing, AMTC trace tracking, and FIF decomposition to respiratory-rate and waveform recovery in Wi-Fi systems (Rhodes et al., 31 Jul 2025). A common misconception would be to assume a single unified framework because of the shared term. The evidence instead supports a purely homonymous relation. One sense is an informal domain shorthand tied to WIM magneto-ionic diagnostics, while the other is an explicit algorithm name in wireless health sensing.