Exclusion-Based Telemetry Security

Authentication without
a single biometric sensor.

SentinelSense transforms the mobile sensors your organisation already owns into a continuous, hardware-free security fabric — governing authentication, behavioural monitoring, and physical safety compliance in real time.

Explore the Architecture See the Evidence

50Hz

Telemetry Resolution

Security Domains Covered

Additional Hardware Required

∞

Passive Monitoring Coverage

💳

Traditional Access Control Is Broken

PINs are shared. Cards are cloned. Passwords are phished. Static credentials verify a token, not a person — leaving the door open to impersonation from the moment credentials are issued.

VULNERABILITY

🔬

Hardware Biometrics Are Expensive and Brittle

Fingerprint readers fail in field conditions. Iris scanners require controlled lighting. Deploying dedicated biometric hardware at scale carries a prohibitive capital outlay with ongoing calibration overhead.

COST RISK

📍

Personnel Compliance Is Impossible to Audit

Guards who abandon their post, workers entering restricted zones, or personnel suffering an incapacitating fall — without continuous passive monitoring, these events are invisible until the damage is done.

OPERATIONAL GAP

Platform Capabilities

Three security domains. One telemetry pipeline.

SentinelSense processes raw mobile sensor streams through an exclusion-based model — continuously validating against an authorised multivariate baseline derived from real-world captures.

USE CASE 01

Identity Authentication

Gait-based identity verification that is non-transferable and non-reproducible. The system compares rotation orientation (ry, rz) and environmental context against an authorised multivariate baseline. Peter's dominant gry of ~9.7 m/s² and left-hand hold signature cannot be mimicked — the specific skeletal-muscular biomechanics are unique to each individual.

lax / lay / laz ry / rz light / llight grx / gry / grz

✓ No PIN. No card. No password. Identity is the person's body.

USE CASE 02

Behavioural Firewall

A four-layer verification stack — postural biometrics, environmental integrity, 50Hz micro-tremor vigilance, and geospatial guardrails — ensures the authorised person remains present at the authorised location. Jason's flatter device hold (gry ~5.7–7.0 m/s²) and dim environment (12 lux vs. Peter's 100 lux) trigger failure at Layer 1.

ax / ay lat / lon gry / grz sound

✓ Detects post substitution, zone abandonment, and device handoff instantly.

USE CASE 03

Intelligent Fall Detection

A three-phase algorithm — impact spike detection, orientation shift confirmation, and post-fall immobility monitoring — distinguishes a genuine incapacitating fall from normal movement. 50–100Hz sampling is required to capture the acceleration dynamics of a fall; modern chipsets handle up to 125Hz. Atmospheric cross-validation eliminates false positives.

grx / gry / grz lax / lay / laz press / sound

✓ Lone-worker safety and duty-of-care compliance — automated.

Research Evidence

Every claim is backed by real sensor data.

The following technical infographics were generated directly from live telemetry captures, illustrating how SentinelSense distinguishes authorised users from impostors across multiple sensor dimensions.

Sensor Fingerprinting: Peter vs. Jason

A direct comparison of telemetry from two subjects demonstrates that gravity distribution and ambient light alone are sufficient to uniquely identify an individual — no specialised hardware required.

Peter's gry remains dominant at ~9.7 m/s², reflecting a stable seated posture with left-hand device grip.
Jason's gravity distributes across axes (gry ~5.7, grz ~7.2), revealing a different hold angle entirely.
Ambient light diverges sharply: Peter operates at ~100 lux; Jason at ~12 lux — a near 10× difference that acts as an environmental zone marker.

Biometric Fingerprinting via Mobile Telemetry — Sensor radar comparison of Peter and Jason

Telemetry Signature Analysis

Unique Biometric Profiles via Gravity & Light

The combination of vertical gravity dominance and ambient brightness creates a biometric profile that is as unique as a fingerprint — and far harder to spoof, because it reflects how a person physically inhabits their environment.

Peter's vertical gravity dominance (gry near 9.7 m/s²) reflects highly stable, repeatable left-hand device handling.
Jason's multidimensional device tilt shows a balanced grx/grz relationship — a distinct and immediately detectable signature.
The sensor data is so consistent it constitutes a passive biometric that re-authenticates continuously, not just at login.

Biometric Telemetry User Identification — Gravity sensor bar charts and radar profiles

Behavioural Firewall — In Detail

Four layers. One impostor has nowhere to hide.

The Security Guard Behavioural Firewall processes telemetry through four sequential gates. A failure at any single layer triggers an immediate lockout — making partial spoofing useless.

Security Guard Behavioural Firewall — Four-layer telemetry authentication process

LAYER 1

Postural Biometrics

Gravity vector analysis (gry/grz) flags the impostor's flatter horizontal hold (5.7–7.0 m/s²) against the authorised ~8.7 m/s² vertical-seated profile.

LAYER 2

Environmental Integrity

Ambient light (~100 lux authorised vs. 12 lux impostor) and sound (26–50 dB active zone vs. 0 dB silent profile) must match the assigned guard post's conditions.

LAYER 3

50Hz Micro-Tremor Vigilance

A static or cloned device produces a flat-line gyroscope profile. Humans produce natural high-frequency micro-tremors; their absence is an instant red flag.

LAYER 4

Geospatial Guardrails

GPS must originate within the authorised zone (−33.6422755, 18.4754235). Any significant coordinate deviation triggers an immediate lockout.

Posture Profiling: Sitting vs. Standing

Real telemetry captures across three sessions in June 2026 reveal that even a change in posture generates a distinct, machine-readable gravity signature — enabling the system to detect not just who is holding the device, but how.

Sitting: gry dominates at 8.6–9.0 m/s², grz at 3.6–4.6 m/s². Steep device tilt. High capture density confirms postural stability.
Standing: gravity shifts — gry drops to ~5.8, grz rises to ~7.8 m/s². The gravity component inversion is unambiguous.
Left-hand consistency across both postures confirms Peter's stable grip pattern, adding a further exclusion dimension against impostors who hold differently.

Posture Profiling Peter Sensor Data Analysis — Sitting vs Standing gravity components

Fall Detection: Three Phases to Certainty

Falls require 50–100Hz sensor sampling to capture accurately. SentinelSense's three-phase detection algorithm uses the full sensor stack to eliminate false positives and ensure genuine emergencies trigger an immediate response.

Phase 1 — Impact Spikes: Massive acceleration spikes across all axes, threshold > 3g, signal a potential fall event.
Phase 2 — Orientation Shift: Gravity sensors confirm the transition from upright to prone — a 90° axis shift is definitive.
Phase 3 — Post-Fall Immobility: Near-zero acceleration variance in the period following impact confirms the user is incapacitated, not simply crouching.

Smartphone Fall Detection Sensor Logic — Three-phase detection with trigger sequence

Technical Architecture

The Exclusion Model

Security is enforced by continuously measuring deviation from an authorised multivariate baseline — not by matching against a stored template.

Baseline Capture

Authorised sensor patterns are captured from the enrolled individual across gravity vectors, linear acceleration, rotation, and environmental channels at 50Hz resolution (Δts ≈ 20ms).

Multivariate Standard Deviation Threshold

A sliding temporal window calculates the standard deviation of lax, lay, and laz in real time, establishing a probabilistic perimeter around the authorised signature.

Continuous Stream Comparison

Incoming telemetry is compared against the threshold on every frame. Orientation inversions (negative rz), light-level deviations, and spatial coordinate drift are caught immediately.

Exclusion Event Dispatch

Any signature deviation beyond the authorised threshold triggers an immediate security flag and escalation — with the specific sensor axes that caused the violation logged for audit.

exclusion_model.pseudo

// Stage 1 — Gravity Baseline Monitoring
MONITOR grz AGAINST baseline [8.6, 8.7]

// Stage 2 — Impact Spike Detection
IF (lax > 10.0 OR lay > 10.0)
  AND (grz < 5.0):

  // Stage 3 — Atmospheric Cross-Validation
  IF (press == 0) AND (sound < 40.0):
    STATUS = "CONTROLLED_MOVEMENT_EXCLUSION"
  ELSE:
    STATUS = "CRITICAL_FALL_TRIGGERED"
    DISPATCH(level=CRITICAL, ts=now)
END IF

// Behavioural Firewall — Spatial Integrity
auth_coord = [-33.9422534, 18.4754618]
IF DELTA(lat, auth_coord[0]) > 0.0000100:
  STATUS = "SPATIAL_INTEGRITY_VIOLATION"
  DISPATCH(level=HIGH, axis=lat)

// Identity — Orientation Exclusion
IF rz < 0 OR ry < 0:
  STATUS = "IDENTITY_EXCLUSION_TRIGGERED"

// Postural Check — Gravity Component
IF gry < 8.0:
  STATUS = "POSTURE_MISMATCH_DETECTED"
      

Data Grounding

Every threshold is empirically derived.

All security boundaries are grounded in real telemetry captures — not arbitrary heuristics. Every range below is extracted directly from authorised-user sensor data.

Security Function	CSV Column(s)	Authorised Range	Violation Indicator	Status
Gait Auth — Lateral Rotation	ry	+0.172 to +0.340	−0.455 (Negative inversion)	IDENTITY GATE
Gait Auth — Vertical Orientation	rz	+0.696 to +0.917	−0.826 (Negative inversion)	IDENTITY GATE
Postural Biometrics — Gravity Y	gry	~9.7 m/s² (seated)	~5.7 m/s² (impostor hold)	POSTURE GATE
Environmental Context — Illuminance	light	98.85 to 115.74 lux	12 lux (dim environment)	ZONE GATE
Gravity Baseline — Fall Pre-condition	grz	8.60 to 8.79 m/s²	< 5.0 m/s² (free-fall)	SAFETY TRIGGER
Impact Detection — Linear Acceleration	lax / lay	−3.41 to −11.50 m/s²	> 10.0 m/s² (impact spike)	FALL TRIGGER
Spatial Integrity — Geographic Perimeter	lat / lon	−33.9422534, 18.4754618	Δ > 0.0000106° (~1.2 m)	ZONE BREACH
Atmospheric Validation — False Positive Filter	press / sound	press = 0, sound < 40 dB	Spike in either confirms impact	VALIDATION LAYER

Why SentinelSense

The competitive case for telemetry-native security.

The sensors are already in every device. The infrastructure exists. SentinelSense is the layer that turns passive hardware into an active, continuous security fabric.

📡

Zero Additional Hardware

Gravity, accelerometer, gyroscope, GPS, light, pressure, and microphone sensors are standard in every modern mobile device. There is nothing to procure, install, or maintain.

🔒

Impersonation Is Structurally Impossible

The harmonic oscillation of an individual's gait, their hold angle, and their ambient environment combine into a profile that cannot be reproduced by borrowing a device or mimicking a walk.

⚡

Continuous, Passive Enforcement

Security is not a checkpoint — it is an unbroken stream. SentinelSense monitors every second of operation at 50Hz, making the gap between authorisation and compromise effectively zero.

🧮

Empirically Grounded Thresholds

Every security boundary is derived from real sensor captures — not vendor specifications or theoretical models. The exclusion model reflects actual behaviour in real operating environments.

Authentication withouta single biometric sensor.