Physics-Based · CC0 Licensed · Peer-Reviewable · v3.2.0

Synthetic IoT Datasets from Physics, not Black Boxes

Every row comes from a documented equation — Bergman glucose ODE, Pareto-tailed DDoS flows, RC thermal relaxation, analytic CO₂ mass balance. Seeded by a portable Mersenne Twister for bit-identical reproducibility, and certified by a Kolmogorov–Smirnov validation report.

Start Free How it Works
6 IoT Domains · Up to 10,000 rows / dataset · 44 peer-reviewed references · KS-tested sidecar JSON · PHP 7.4 – 8.3 bit-identical
v3.2.0

What changed in v3.2

  • Portable MT19937. Pure-PHP Mersenne Twister — same seed produces bit-identical CSV across PHP 7.4, 8.0, 8.1, 8.2, 8.3.
  • Analytic CO₂ ODE. Closed-form solution of the mass-balance equation; no Euler drift.
  • RC thermal relaxation. ISO 13790 single-zone model with τ = R·C; replaces the v3.1 direct Fourier-as-indoor assumption.
  • Psychrometric humidity. Magnus–Tetens saturation formula replaces the Cholesky correlation.
  • Heavy-tailed DDoS. Pareto flow durations and packet counts (α=1.2 DDoS, 1.3 DoS) per Crovella & Bestavros (1997) and Antonakakis et al. (2017).
  • Bergman Minimal Model. 3-state glucose–insulin ODE (RK4, 1-min substeps) with Dalla Man (2007) meal absorption; diabetic variant included.
  • Validation Layer. Every CSV now ships with a *.validation.json sidecar: descriptive stats, KS tests, physical invariants, signed correlations.
  • AI-discovery metadata. llms.txt, llms-full.txt, Google-Dataset-Search JSON-LD, and dynamic OG preview cards.
🏠

Smart Home

RC thermal relaxation (ISO 13790), analytic CO₂ ODE, Magnus–Tetens psychrometric humidity, Markov occupancy, deadband HVAC

🛡️

IoT Security

LogNormal benign packets, 4-phase Markov attack bursts, Pareto-tailed DoS/DDoS flows (Crovella & Bestavros; Antonakakis et al.)

⚙️

Predictive Maintenance

Weibull degradation, ISO 10816 vibration zones, RUL-ready bearing temperature and current models

🏥

Medical IoT

Circadian HR/BP/SpO₂, NEWS2 scoring, Bergman Minimal Model (RK4) with Dalla Man (2007) meal absorption

🏭

IIoT Network

Modbus / OPC UA / DNP3 traffic, OT roles (PLC, HMI, SCADA, RTU), MitM / replay / false-data-injection attacks

🚗

Connected Vehicle

Driving state machine, GPS dead reckoning, 5-gear RPM model, event classifier (hard-brake / rapid-accel)

Generate a Dataset

Choose domain, configure parameters, download CSV.

Ready to Generate?

Create a free account to start generating research-grade IoT datasets. Your first generation is on us.

Create Free Account Sign In

1 free generation · No credit card required

How IoTSyn Works

Transparent, reproducible data generation grounded in established mathematical frameworks.

1

Choose Domain & Parameters

Select from 6 IoT domains. Configure physical parameters — climate, equipment type, patient demographics, network topology. Defaults are calibrated from literature.

2

Physics-Based Generation

Data is generated from explicit mathematical models — Fourier decomposition, mass-balance ODEs, Weibull distributions, Markov chains. Every equation is documented.

3

Download & Cite

Download as CSV with metadata header. Each dataset includes its seed for exact reproduction, and auto-generated citations in APA, MLA, Chicago, IEEE, BibTeX, and Harvard.

Sample Mathematical Models (v3.2.0)

Indoor temperature — RC thermal relaxation (ISO 13790)

T(t+Δt) = T_target + (T(t) − T_target)·exp(−Δt/τ) + ε_AR(1)

CO₂ — analytic solution of mass-balance ODE

C(t+Δt) = C_ss + (C(t) − C_ss)·exp(−Δt/τ_vent), C_ss = C_out + n·G/Q

Glucose — Bergman Minimal Model (Bergman et al., 1979)

dG/dt = −(p₁ + X)·G + p₁·G_b + D(t); dX/dt = −p₂·X + p₃·(I − I_b)

DDoS flow duration — Pareto heavy-tail (α=1.2)

P(X > x) = (x_m / x)^α, X = x_m · (1 − U)^(−1/α)

Equipment degradation — Weibull CDF (ISO 10816)

D(t) = 1 − exp(−(t/L)^β)

Validation — Kolmogorov–Smirnov goodness-of-fit

D_n = sup_x |F_n(x) − F(x; θ̂)|; P(√n·D_n > t) = 2·Σ (−1)^(k−1)·exp(−2k²t²)
See complete methodology with all equations →

📄 Technical Report

Full mathematical specification of all 6 generators, with 10 academic references including Box-Muller, Knuth, Marsaglia-Tsang, and ISO 10816.

Read & cite the technical report →

📦 Public Dataset Repository

Generated datasets are periodically published to IoTDataset.com for direct browsing and download.

Visit IoTDataset.com →