Table 2 Comparison of AI-enhanced photonic nose with other gas sensing technologies

Selectivity

High—uses optical spectral fingerprint or multiple specific optical channels; AI decodes mixtures.

Medium—relies on overlapping responses and AI; cross-sensitivity common.

Very High—physically separates and identifies each compound.

Sensitivity

High—sub-ppm routinely, ppb with enhancements; optical cavity methods to boost signals.

Medium—ppm typically; some nanomaterial sensors reach ppb but with noise.

Very-High–ppb to ppt for many compounds with enrichment.

Response Time

Fast—often <1 s (limited by sampling); real-time readout.

Moderate—seconds to minutes (sensor response and recovery time, esp. if heated).

Slow—minutes for chromatographic separation and analysis.

Drift /Stability

Good—optical components stable; minimal baseline drift. Periodic calibration may be needed for coatings.

Poor to Moderate—significant drift over time; frequent recalibration or baseline correction required.

Excellent per run; instrument itself needs maintenance/calibration over long term.

Power Consumption

Moderate—needs optical source and photodetectors; can be optimized for low power (edge use).

Low to Moderate—MOX sensors need heaters; others low-power; MCU usage minimal.

Very High—lab equipment with vacuum pumps, ovens, detectors, and a PC.

Size/Portability

Small—chip-scale sensors; with integrated optics can be palm-sized device or smaller.

Small—e-noses can be handheld or wearable easily.

Large—benchtop or larger, not portable in field conditions.

Data Complexity

High—rich data requiring AI interpretation.

High—multivariate data requiring AI, but typically fewer channels than photonic.

High—but processed into human-readable format by software.

Deployment Cost

Emerging—currently higher than simple gas sensors, but potential for cost reduction via mass production.

Low—many cheap sensors available; total system cost is low.

Very High—expensive instrument and skilled operators needed.

Notable Advantages

Multi-analyte detection, high specificity without reagents, fast and AI-adaptive.

Simplicity, low cost, broad response (one device can respond to many chemicals).

Definitive identification and quantification.

Notable Limitations

Still developing; might require sophisticated calibration; some designs sensitive to environmental conditions.

Prone to drift and false positives; often requires frequent recalibration; limited ability to identify specific compounds without ambiguity.

Not continuous monitoring; not suitable for on-site or real-time alerts; high maintenance.