Methodology and validation

IoTSyn Smart City & Public Spaces v4.4.0

A specification-driven synthetic urban-data generator and decision-layer testbed. This page documents the v4.4.0 Smart City reference engine first, then identifies the five earlier framework domains separately so their provenance and validation claims are not conflated.

Smart City reference engine v4.4.0 Legacy framework v3.2.0 Public research artifacts Internal verification, not site validation
Frozen public record

The public record includes data, results, and a reference implementation

The v4.4.0 research record contains scenarios, verified outputs, contracts, vectors, and external-comparison data. The RC1 specification bundle also contains a pure-PHP reference implementation under CC0. Private production-service code, deployment configuration, credentials, and databases remain outside the public bundle.

Cite the frozen v4.4.0 record → Read the implementation specification →

Contents

1. Scope and version boundary 2. Smart City v4.4.0 method 3. Verification and reproducibility 4. Legacy framework domains 5. Scope and limitations 6. References

1. Scope and version boundary

IoTSyn generates synthetic data from declared physical, stochastic, and decision models; it does not infer unknown latent patterns from real proprietary datasets. The current website contains two explicitly different methodological tracks.

v4.4.0: Smart City & Public Spaces

The reference engine is a normative 22-stage pipeline for a hypothetical urban public space. It defines data layers, random-consumption order, artifact contracts, and verification procedures for coupled environment, presence, sensor-observation, and decision outputs.

Legacy framework: v3.2.0

Smart Home, Predictive Maintenance, Medical IoT, IIoT Network Traffic, and Connected Vehicle generators remain documented framework domains. Their models and validation summaries are not relabelled as v4.4.0 Smart City evidence.

Claim boundary

The v4.4.0 evidence supports byte-exact Tier 1 reproduction in the tested reference environment, internal implementation verification, and synthetic decision sensitivity under declared observation-error models. It does not establish site validation, cross-site predictive skill, unrestricted cross-machine identity, or identity between the public reference implementation and the private production service.

2. Smart City & Public Spaces v4.4.0 method

The method separates latent physical truth, sensor observation, aggregation/features, and risk/decision output. This prevents an observed sensor signal from being mistaken for the simulated physical state that generated it.

Layer Role Examples
Latent physical truthDeclared environmental and crowd stateMeteorology, pollutant concentration, occupancy
Sensor observationSynthetic measurement processBias, drift, noise, missingness
Aggregation and featuresWindowed or derived quantities24-hour PM2.5, LAeq window, crowding ratio
Risk and decisionOperational classifications and guidanceHazard levels, action level, episodes, strategy objects

2.1 Pipeline and physical-stochastic components

Time, solar geometry, and meteorology

The pipeline fixes time-grid semantics, calendar, location, and named random substreams. The reference implementation uses MT19937 and pinned sampler behaviour [13] [14] [15]. Solar position uses the versioned iotsyn_solar_v1 module based on the NOAA approximation equations and Meeus [1]. Cloud transmissivity follows Kasten and Czeplak [2]; vapour pressure uses the WMO-recommended Sonntag coefficients rather than treating historical Tetens values as the v4.4 source [3] [4].

Thermal exposure and presence

UTCI is the primary year-round outdoor thermal index; WBGT is an optional heat-only path with separate policy metadata [5] [6]. Pedestrian arrivals are a discrete-time, state-modulated NHPP; capacity, admission, dwell time, and occupancy are represented as distinct processes [7] [8].

Air quality and acoustics

PM2.5 and NO2 are generated through a declared box-model pathway and then observed through a separate sensor model. The operational air-quality bands are simulation thresholds inspired by the WHO 2021 guideline and interim-target structure; they are not clinical alert thresholds [9]. Environmental noise is composed in the energy domain over a stated averaging window [10].

Risk and decision layer

Thermal, air-quality, noise, and crowding components form an explicit risk vector. The engine records completeness, overall action level, elevated-hazard count, risk episodes, and provenance tags for any behaviour-change strategy object. Decision objects identify whether they are nowcasts, forecasts, or ex-post planning outputs.

2.2 Public artifacts and contracts

The public v4.4.0 research-artifacts record contains a scenario suite, verified outputs, machine-readable CSV/data contracts, schemas, tolerance information, and reference vectors. Canonical JSON uses RFC 8785 [11]. Each run records the configuration and provenance needed to inspect the generated result without exposing implementation source.

3. Verification and reproducibility

Verification asks whether the reference implementation executes the declared specification. It is distinct from external validation, which would ask whether a configuration represents a particular real public space.

What is checked

  • Known-answer vectors for random streams, distributions, solar geometry, physical calculations, and risk logic.
  • Data-contract and schema conformance, physical invariants, and directional scenario checks.
  • Distributional checks use the Kolmogorov-Smirnov test only for designated independent components, not for final autocorrelated series [20].
  • A dedicated offered-NHPP time-rescaling experiment using exact event offsets [12].
  • Reference-scenario outputs and manifest-level provenance.

What is not established

  • The time-rescaling experiment does not validate state-modulated or capacity-managed arrivals.
  • The Melbourne diagnostic is a negative held-out diagnostic, not evidence of cross-site transfer.
  • No site calibration, cross-site predictive skill, or real-world pedestrian-entry model is claimed.
  • The public Zenodo record is not a source-code release.

3.1 Reproducibility contract

Tier 1: verified

The frozen reference run is byte-exact in the tested reference environment. Decimal master_seed, named substreams, resolved configuration, and artifact contracts are part of that verification boundary.

Tier 2: specified, not yet claimed

Cross-runtime and cross-machine reproduction are specified but not asserted as verified. The page therefore does not make a general byte-identical claim across hosts or runtimes.

4. Legacy framework domains (v3.2.0)

The five domains below remain part of the public platform. They are presented as documented legacy framework generators, not as v4.4.0 Smart City components or validation evidence.

Domain Declared model family Boundary
Smart HomeRC thermal relaxation, analytic CO2 mass balance, humidity, occupancy states [17]A simplified single-zone model; not a current ISO compliance claim.
Predictive MaintenanceWeibull degradation and vibration/thermal loadingHistorical ISO 10816 context only [22]; the standard is withdrawn and not silently replaced.
Medical IoTBergman glucose model and meal absorption [18] [19]Synthetic physiology for research and education; not clinical decision support.
IIoT Network TrafficProtocol-labelled traffic and attack scenariosDoes not reproduce protocol-level artifacts of a specific botnet or deployment.
Connected VehicleDriving states, telemetry, and fuel modelA synthetic behavioural model, not a calibrated fleet model.

ISO 13790:2008 and ISO 10816-1:1995 remain historical provenance for inherited approximations [21] [22]; both are withdrawn. ISO 52016-1:2017 and ISO 20816-1:2016 are current replacements [23] [24], but they are not claimed as implemented unless the corresponding models are revised and validated against them.

5. Scope and limitations

  • Physics-based structure and internal verification do not establish correspondence to every real-world environment.
  • UTCI, air-quality, noise, and crowding decision outputs are declared operational simulation constructs. They are not clinical, legal, or municipal alert systems.
  • The v4.4.0 Melbourne diagnostic provides no evidence of external transfer; a future benchmark must use a frozen data snapshot, eligibility gate, named calibration controls, and separate calibration/evaluation windows.
  • Generated datasets may be CC0 where their output metadata says so; the proprietary platform implementation is not released under CC0.

6. References

Only references used by this page are listed. The full normative source list remains in the v4.4.0 specification and manuscript.

Smart City v4.4.0

  1. Meeus, J. (1998). Astronomical Algorithms, 2nd ed. Willmann-Bell.
  2. Kasten, F., & Czeplak, G. (1980). Solar and terrestrial radiation dependent on the amount and type of cloud. Solar Energy, 24(2), 177–189.
  3. Sonntag, D. (1990). Important new values of the physical constants of 1986, vapour pressure formulations based on the ITS-90, and psychrometer formulae. Zeitschrift für Meteorologie, 40(5), 340–344.
  4. World Meteorological Organization. (2018). Guide to Instruments and Methods of Observation (WMO-No. 8).
  5. Bröde, P., et al. (2012). Deriving the operational procedure for the Universal Thermal Climate Index (UTCI). International Journal of Biometeorology, 56(3), 481–494.
  6. International Organization for Standardization. (2017). ISO 7243: Ergonomics of the thermal environment - Assessment of heat stress using the WBGT index.
  7. Fischer, W., & Meier-Hellstern, K. (1993). The Markov-modulated Poisson process cookbook. Performance Evaluation, 18(2), 149–171.
  8. Lewis, P.A.W., & Shedler, G.S. (1979). Simulation of nonhomogeneous Poisson processes by thinning. Naval Research Logistics Quarterly, 26(3), 403–413.
  9. World Health Organization. (2021). WHO Global Air Quality Guidelines: Particulate Matter (PM2.5 and PM10), Ozone, Nitrogen Dioxide, Sulfur Dioxide and Carbon Monoxide.
  10. International Organization for Standardization. (2016). ISO 1996-1: Acoustics - Description, measurement and assessment of environmental noise - Part 1.
  11. Rundgren, A., Jordan, B., & Erdtman, S. (2020). JSON Canonicalization Scheme (JCS). RFC 8785.
  12. Brown, E.N., Barbieri, R., Ventura, V., Kass, R.E., & Frank, L.M. (2002). The time-rescaling theorem and its application to neural spike train data analysis. Neural Computation, 14(2), 325–346.
  13. Matsumoto, M., & Nishimura, T. (1998). Mersenne Twister: a 623-dimensionally equidistributed uniform pseudo-random number generator. ACM Transactions on Modeling and Computer Simulation, 8(1), 3–30.
  14. Box, G.E.P., & Muller, M.E. (1958). A note on the generation of random normal deviates. Annals of Mathematical Statistics, 29(2), 610–611.
  15. Marsaglia, G., & Tsang, W.W. (2000). A simple method for generating gamma variables. ACM Transactions on Modeling and Computer Simulation, 26(3), 363–372.
  16. Duhair, A. (2026). IoTSyn v4.4.0 public research artifacts: Scenarios, verified outputs, contracts and external-comparison data [Data set]. Zenodo. https://doi.org/10.5281/zenodo.21480873.

Legacy framework domains

  1. Persily, A.K., & de Jonge, L. (2017). Carbon dioxide generation rates for building occupants. Indoor Air, 27(5), 868–879.
  2. Bergman, R.N., Ider, Y.Z., Bowden, C.R., & Cobelli, C. (1979). Quantitative estimation of insulin sensitivity. American Journal of Physiology, 236(6), E667–E677.
  3. Dalla Man, C., Rizza, R.A., & Cobelli, C. (2007). Meal simulation model of the glucose-insulin system. IEEE Transactions on Biomedical Engineering, 54(10), 1740–1749.
  4. Kolmogorov, A.N. (1933). Sulla determinazione empirica di una legge di distribuzione. Giornale dell'Istituto Italiano degli Attuari, 4, 83–91.
  5. ISO 13790:2008. Energy performance of buildings - Calculation of energy use for space heating and cooling. Withdrawn; retained here only as historical provenance.
  6. ISO 10816-1:1995. Mechanical vibration - Evaluation of machine vibration by measurements on non-rotating parts. Withdrawn; retained here only as historical provenance.
  7. ISO 52016-1:2017. Energy performance of buildings - Energy needs for heating and cooling, internal temperatures and sensible and latent heat loads - Part 1: Calculation procedures. Current successor to ISO 13790:2008; listed for context, not as an implementation-compliance claim.
  8. ISO 20816-1:2016. Mechanical vibration - Measurement and evaluation of machine vibration - Part 1: General guidelines. Current successor to ISO 10816-1:1995; listed for context, not as an implementation-compliance claim.

Use the right citation for the material you used

Cite individual datasets from their view pages. Cite the v4.4.0 Zenodo record for the Smart City research artifacts, and cite the platform separately only when it is the object of discussion.

Citation guidance Explore Smart City datasets