Files
paper-how-to-use-this-techn…/main.md
T

24 KiB
Raw Blame History

PulseCast: Turning a Noisy Footfall Index into a Weekly Staffing Plan

Authors: [AUTHOR TODO]
Affiliation: [AFFILIATION TODO]
Date: 14 November 2026
Artifact: [demo URL TODO] · [code repository TODO]

Abstract

Inner-city store managers must decide next week's shifts without knowing next week's footfall. We present PulseCast, a decision-first staffing copilot built on eight passive Bluetooth footfall counters in a ~580 m × 280 m cluster in Heilbronn, Germany. The system fuses five months of 15-minute device counts with Open-Meteo weather and Baden-Württemberg public holidays, trains a global gradient-boosted model on a time-based split, and converts the forecast into a weekly shift plan through a transparent staffing heuristic. On a 28-day holdout, the weather-aware 7-day model reaches MAE 3.32, WAPE 27.5%, and R² 0.777, versus MAE 5.40, WAPE 44.7%, and R² 0.462 for a seasonal-naive baseline. The plan is generated in under two seconds and the system is deployed on STACKIT SKE behind a Cloudflare Tunnel. A code-level audit shows that each published value is a ~10-second count of randomized BLE addresses multiplied by 0.7, not a 15-minute count of people. We therefore frame the result as a relative activity index that supports staffing decisions, not as a city-wide people counter, and we state what calibration and privacy work would be required before absolute claims.

1. Introduction

A store manager on Kaiserstraße knows the weather forecast and the staff roster. She does not know how many people will walk past next Saturday. Raw counts are not a decision; the manager needs a forecast, an explicit staffing rule, and a way to test a scenario. A dashboard that shows history does not close that gap.

The Schwarz Digits Pulse of the City challenge asks teams to turn privacy-preserving pedestrian counts from Heilbronn into predictive tooling for advertising, store locations, and staffing [1]. We chose staffing: eight fixed locations, five months of multi-sensor history, and a clear daily and weekly rhythm. Store-location ranking would need city-wide coverage that eight sensors in a 580 m cluster cannot provide; a city-wide heatmap would be a visualization, not a measurement.

PulseCast makes three contributions. It treats the decision as the product: the forecast feeds a weekly shift plan. It uses a global, weather-aware gradient-boosted model with a time-based holdout and a seasonal-naive baseline, rather than a model the data cannot support. It is deployed as a three-service system on SKE with a cached fallback and a live scenario layer. It also reports a code-level audit showing that the published value is a short-window count of randomized BLE addresses scaled by 0.7, not a 15-minute count of people. That audit changes how the results should be read and what the next version must fix.

We do not claim a production people counter, a city-wide heatmap, annual seasonality, live sensor data, or a proven return on investment. We claim a working decision loop, a credible relative forecast, and an honest account of the measurement limits.

2. Data

2.1 The delivered PAX data

The dataset contains 192,464 CSV rows: 96,232 density readings and 96,232 battery readings. There are 12 device IDs; after merging the two density-01 device IDs and excluding three test sensors, eight usable locations remain in a ~580 m × 280 m cluster around Hagenbuchersee, Neckarufer, Stadtgalerie, Marahaus and Götzenturm. The eight-sensor window runs from 12 June to 7 November 2026, with a median cadence of 15.3 minutes. Values range from 0 to 292.6; the multi-sensor mean is ~14.4, the median 7.7, p95 46.9 and p99 78.4. The inner city peaks between 18:00 and 21:00 local time, with the top at 19:00; Saturday is the busiest day on average (mean 19.1) [2].

The sensors are LoRaWAN-connected PAX counters that passively sniff Bluetooth Low Energy (BLE) advertisements and transmit aggregated device counts [1, 2]. The challenge brief calls them privacy-preserving because they count devices rather than identify people. The counts are estimates, not a census: device penetration, Bluetooth settings, and the scanning design all matter.

2.2 What one value actually is

A code-level audit of the deployed Code for Heilbronn lorax firmware changes the interpretation of every number in this paper [3, 4]. The node wakes from deep sleep, scans BLE passively for about 10 seconds, counts distinct locally administered (“randomized”) addresses, truncates the last two address bytes into a 16-bit ID, multiplies the integer count by a per-sensor factor (default 0.7), sends one LoRaWAN payload, and sleeps for 900 seconds [3, 4].

The published value is therefore 0.7 × U, where U is the number of distinct randomized BLE addresses seen in a ~10-second scan. The 15-minute cadence is the sleep cycle, not the integration window. The effective duty cycle is about 1.1%. The firmware sets a 30-second scan duration, but the released code stops on the first 10-second report callback; the 30-second value is not enforced by the scanner [4].

The audit is of the public 2026 firmware; the deployed 2025 binary may differ, but the data are consistent with the 0.7 scaling. Every one of the 83,156 density values in the eight-sensor panel is an exact multiple of 0.7. The reconstructed integer count has median 12, p95 71, and a maximum of 418; 2.7% of readings are zero. The legacy density-01 device reports integers on a 1.0 grid until March 2026, then a replacement device reports on the 0.7 grid from April 2026, with a 35.2-day gap between the two series. The two density-01 series are not on the same scale and must not be pooled [5].

This is a relative activity index, not a count of people. A person who passes in 30 seconds has roughly a 4.4% chance of overlapping a ~10-second scan; a person who lingers for an hour can be counted in up to four consecutive bins. The index conflates flow and dwell, and it cannot distinguish a phone from a smartwatch, earbud case, beacon, or car head-unit. Device penetration and the people-per-device ratio are unknown. The 0.7 factor is a provisioning default, not a validated penetration correction.

Practical consequence. The model predicts the published index well enough to compare relative peaks and to drive a staffing rule. It does not predict people, and it does not support absolute footfall or revenue claims without a calibration study.

2.3 Coverage, gaps, and context

Coverage is uneven. Only 28.2% of 15-minute slots in the analysis panel have all eight sensors reporting; the modal count is seven. Aggregating to 30, 60, or 120 minutes raises complete eight-sensor coverage to 63.4%, 69.7%, and 71.0%, respectively [5]. About 5.2% of slots have all sensors at zero and 39.6% have all sensors below 5. The worst month falls to 23.3% coverage at one location. Gaps reach four to six days per sensor; the cause is undocumented, but the missing periods differ systematically from full-coverage periods [2, 5].

Weather comes from Open-Meteo: the historical archive for training and validation, and the current/7-day forecast for the dashboard and what-if scenarios. Temperature and precipitation are the useful model features; cloud cover and wind did not improve the short test. Public holidays for Baden-Württemberg are included as an is_holiday feature and as dashboard badges. They added little in the 28-day test and should be read as context, not as a large accuracy driver [2].

Sensor Readings Mean Median p95 Max Max daily gap
hn-stadtgalerie-01 8,982 21.1 8.4 67.2 137.2 6 days
hn-neckarufer-01 9,527 19.2 11.9 61.6 255.5 5 days
hn-hagenbuchersee-02 10,099 19.3 11.2 64.4 200.9 4 days
hn-marahaus-01 9,254 13.7 9.1 38.5 194.6 6 days
hn-hagenbuchersee-01 12,345 13.0 6.3 49.7 292.6 4 days
hn-goetzentuem-01 9,952 11.1 8.4 29.4 210.7 5 days
hn-soleo-01 10,327 10.6 8.4 28.0 111.3 4 days
density-01 12,670 9.9 7.7 28.0 63.0 35 days

Table 1. Data coverage per sensor in the eight-sensor network. density-01 uses the longer history; the 35-day gap is across the hardware replacement.

Figure 1. Sensor map.

Figure 1. The eight sensor locations in the ~580 m × 280 m inner-city cluster. Measured points only; no city-wide surface.

Figure 2. Hourly and weekday profile.

Figure 2. The “pulse”: mean index by hour and weekday across the eight sensors. The peak is 18:00–21:00 local; Saturday is the busiest day on average.

3. Method

3.1 Preprocessing and features

We parse timestamps as UTC, convert to Europe/Berlin for calendar features, and handle daylight saving explicitly. We treat the two density-01 device IDs as one logical location for the eight-sensor panel, but we do not pool the legacy integer series with the 0.7-scaled replacement: the model’s training window uses the current device only. We exclude the three test sensors (maker-fair, niklas-01, hn-bollwerksturm-01), aggregate to hourly means per sensor, and keep the 15-minute data for the drill-down view. The target is the hourly mean of the published index, not a people count.

Features are calendar (hour, weekday, month, weekend, holiday), lags (24, 48, 168 hours, plus 24-hour and 7-day rolling means), temperature and precipitation from Open-Meteo, and sensor identity so the global model can learn location-specific levels. All lags and rolling windows end at the forecast issue time; the time-based split prevents leakage across the holdout.

3.2 Model and validation

We train a global HistGradientBoostingRegressor across all eight sensors. The training window is 12 June to 10 October 2026; the test period is the last 28 days, 10 October to 7 November 2026. We compare against a seasonal-naive baseline that predicts the same hour one week earlier, and report MAE, WAPE, and R². The global model pools strength across sensors while keeping sensor identity as a feature; with eight series and five months, a deep sequence model is not justified and would be harder to validate.

The model serves four horizons: 15 minutes, 1 hour, 48 hours, and 7 days. The 15-minute and 1-hour horizons are nowcast/replay because the dataset is a static CSV and there is no live sensor feed. The 48-hour and 7-day horizons are the planning horizons; the 7-day forecast is the hero.

3.3 From forecast to shift plan

The ML model predicts footfall. A separate, deterministic heuristic turns the forecast into staff:

forecast footfall per hour
  → visitors per staff per hour (service standard)
  → required staff = ceil(footfall / standard × safety factor)
  → minimum staff, shift length, opening hours
  → shift blocks + coverage score + estimated cost

The assumptions are visible and adjustable: visitors per staff hour, labor cost per hour, minimum staff, shift length, and safety factor. The manager can accept, edit, or reject the plan and regenerate it with a scenario. This is decision support with explicit assumptions, not an optimizer with proven ROI. We do not have POS, sales, queue, or labor-cost data, and we do not claim a precise revenue or productivity effect.

4. System

PulseCast is deployed as three namespace-scoped Deployments on STACKIT SKE plus a Cloudflare Tunnel connector: Nuxt 4 SSR + Onyx for the server-rendered shell, map, charts, forecast, plan, and scenario controls; a Go API for sensor/series/forecast/staffing endpoints, SSE fan-out, scenario state, and a cached fallback; and a FastAPI ML service for /predict and /simulate, loading a model baked into the image at build time.

The namespace does not allow Ingress, Jobs, CronJobs, cert-manager, or external-dns. We therefore expose the app through a Cloudflare Tunnel, terminate TLS at Cloudflare, and run the batch pipeline at build time. The model and the fallback JSON are baked into the images; there is no database or PVC. If FastAPI is unavailable, the Go API serves the cached forecast and labels it as cached in the UI. SSE carries scenario changes with a 15-second heartbeat; polling is the fallback. The plan is generated in under two seconds.

The system runs as non-root with resource limits and a read-only root filesystem. The public dashboard is read-only; /control uses basic auth. We do not store PII or MAC addresses in the application. The deployment is a hackathon deployment: one API replica, no SSO, no HA, no live ingestion. We document those gaps rather than hiding them.

Figure 3. System architecture.

Figure 3. Three services plus cloudflared: Nuxt SSR, Go API/SSE and cache, FastAPI ML. Build-time batch; cached fallback; no Ingress, cert-manager, or Jobs.

Figure 4. Plan screen.

Figure 4. The plan screen: recommended shift blocks, coverage, estimated cost, and the “why this plan” context card. The plan is the product; the map and charts support it.

5. Evaluation

5.1 Forecast accuracy

Table 2 reports the 28-day holdout from the generated metrics.json. The weather-aware 7-day model reduces MAE by about 19% relative to the no-weather 7-day model (4.08 → 3.32) and by about 39% relative to seasonal naive (5.40 → 3.32). WAPE falls from 44.7% to 27.5%, and R² rises from 0.462 to 0.777. The day-ahead weather model reaches MAE 3.36, WAPE 27.8%, and R² 0.763; the longer horizon is slightly weaker but remains well above the baseline.

Model MAE WAPE R²
Seasonal naive (same hour last week) 5.40 44.7% 0.462
Day-ahead, no weather 3.96 32.8% 0.706
Day-ahead, + temperature/rain 3.36 27.8% 0.763
7-day-ahead, no weather 4.08 33.8% 0.697
7-day-ahead, + temperature/rain (production) 3.32 27.5% 0.777

Table 2. Model comparison on the last 28 days, from the generated metrics.json (n = 5,350). The mission brief lists slightly different earlier numbers (3.36/27.8/0.779); the generated metrics are the source of truth and must be confirmed by the human before submission. The target is the published footfall index, not people.

Figure 5. Forecast versus actual.

Figure 5. Forecast versus actual for one sensor over the 28-day holdout, with an uncertainty band and weather/holiday context markers. The band is a planning aid, not a calibrated 95% interval unless the coverage is reported.

5.2 Ablation and per-sensor behavior

Weather is the main accuracy driver. In the day-ahead ablation, removing it raises MAE from 3.36 to 3.96. The holiday flag changes MAE from 3.96 to 3.84, a 0.12 improvement; it is a small feature and should be read as context, not as a large lift. The production 7-day model is 3.32 with weather and 4.08 without. These are single-site, 28-day results; they are credible as a relative signal, not as a general weather effect.

Per-sensor errors vary with location and coverage (Table 3). The global model is not a claim that all sensors behave the same: sensor identity remains a feature, and the local level differences are real. Stadtgalerie has the highest MAE (5.83) and Hagenbuchersee-02 the highest WAPE (31.1%); density-01 has the lowest MAE (2.26). The variation is consistent with local differences in level, volatility, and coverage, not with a single homogeneous field.

Sensor MAE WAPE R²
density-01 2.26 30.8% 0.722
hn-goetzentuem-01 2.40 26.9% 0.718
hn-hagenbuchersee-01 2.47 38.6% 0.602
hn-marahaus-01 2.67 29.8% 0.671
hn-soleo-01 3.19 20.2% 0.548
hn-neckarufer-01 3.63 28.2% 0.695
hn-hagenbuchersee-02 4.12 31.1% 0.731
hn-stadtgalerie-01 5.83 25.3% 0.765

Table 3. Per-sensor holdout error for the production 7-day model (n ≈ 669 per sensor). The global model is pooled; the local errors are not.

5.3 System metrics

Metric Value
Plan generation latency < 2 s
Concurrent SSE clients [TODO: k6 result]
Web/API/ML image sizes [TODO]
Cold deployment time [TODO]

Table 4. System metrics. We report only measured values; [TODO] items must be filled from the real run before submission.

A local 3-VU validation of the SSE path opened and closed all three connections cleanly, with a p95 event latency of 656 ms. That is a smoke test, not the 120-client requirement; the cluster run is still [TODO] and must not be claimed until it is measured.

The plan screen compares the recommended schedule to a flat schedule on coverage and estimated cost under the stated assumptions. This is a heuristic comparison, not an ROI claim: we do not have actual labor cost, sales, queue, or conversion data.

5.4 What the evaluation does and does not show

The evaluation shows that a global, weather-aware model beats a seasonal-naive baseline on a time-based holdout in one inner-city cluster. It does not show that the model predicts people, that the result generalizes to another city, that the weather effect is causal, or that the staffing plan improves profit. The 28-day test is short, the weather is observed rather than forecast, and the target is a ~10-second BLE index. The result is credible as a relative decision-support signal, not as a breakthrough.

6. Discussion and Limitations

The main limitation is the measurement, not the model. The sensor samples about 10 seconds out of every 15 minutes, counts only randomized BLE addresses, truncates the address to 16 bits, multiplies by a fixed 0.7 factor, and cannot distinguish people from devices. The published value is a useful relative index, but it is not a census and not a 15-minute count. Device penetration is unknown. A single global factor cannot absorb device ownership, multiple devices per person, Bluetooth settings, and the short scan window.

Coverage is also a limitation. Eight sensors in a 580 m × 280 m cluster cannot support a city-wide heatmap or a store-location recommender. Only 28.2% of 15-minute slots have all eight sensors, and gaps are not random. The 28-day holdout is short, the weather is observed rather than forecast, and there is no event log, POS data, or labor-cost data. The deployment has one API replica, no SSO, no HA, and no live sensor ingestion. The 15-minute and 1-hour horizons are replay/nowcast, not a live feed.

The privacy claim also needs care. The application stores no MAC addresses, and only aggregate counts leave the sensor. But the sensing stage still processes link-layer identifiers at the edge; the 16-bit ID is a pseudonymous dedup key, not a cryptographic hash; and passive MAC sensing in public space has a contested legal basis in Germany. The “privacy-preserving” label is an architectural claim, not a formal guarantee. A production deployment would need a DPIA, a documented legal basis, data minimization, retention limits, and, where possible, a sensor that does not extract MAC addresses at all.

What we would do with more resources. Calibrate with manual or camera counts at each site; change the sensing layer to thermal arrays at narrow entrances, edge cameras as reference, mmWave at wide crossings, and aggregated probe data for city context; use forecast weather and a structured event calendar; add POS/queue/labor data and evaluate at the decision level; and replace the single API replica with an authenticated, observable, multi-tenant service. The order matters: fix the measurement before scaling the model.

Passive Bluetooth and Wi-Fi sensing has a long history in pedestrian counting. Studies consistently find that detection is a small, biased, and variable sample of people; counts track relative trends, but absolute levels require local calibration [6, 7]. MAC address randomization reduces linkability but does not make passive counting a census, and it can be defeated by fingerprinting in some settings [7]. Our audit is consistent with this literature and makes the sampling design explicit.

For forecasting, gradient-boosted trees remain a strong baseline for tabular, covariate-rich, small-data problems [8]. Our global model follows that choice. Weather is a known driver of pedestrian activity, but its magnitude is context-dependent [9]. We treat weather as a predictive feature, not a causal effect. Smart-city dashboards often stop at visualization; we focus on the decision layer and on the measurement limits that determine whether a dashboard is trustworthy.

8. Conclusion

PulseCast shows that a small, weather-aware global model can turn a noisy inner-city footfall index into a weekly staffing plan that beats a seasonal-naive baseline and runs in under two seconds on a real deployment. It also shows that the index is not a people counter: each value is a short BLE scan scaled by 0.7, sampled every 15 minutes, with unknown penetration and incomplete coverage. The honest conclusion is narrower and more useful than a city-scale claim: the decision loop works, the relative forecast is credible, and the next step is calibration and a privacy-first sensing layer. We would rather ship a plan a manager can use and state what it does not measure than promise a city-wide heatmap the data cannot support.

Artifacts, data, and AI disclosure

The live demo is at [TODO: URL]; the code is at [TODO: repository]. The raw challenge CSV is not redistributed. Derived aggregates, the trained model card, and the metrics JSON can be shared subject to the organizers’ permission. The application stores no MAC addresses or PII; the sensing-stage limitations and the privacy/legal caveats are described in Sections 2.2 and 6. A production deployment would require a DPIA and a documented legal basis before collecting MAC-derived counts.

AI tools were used for drafting and editing. The authors verified every number, citation, and claim against the delivered data, the project metrics, or a cited source. AI tools are not authors and take no responsibility for the content.

References

[1] Schwarz Digits / Code for Heilbronn. The Pulse of the City, challenge brief, 2026. Local: OKF_Vault/Schwarz Digits/FCH Challenge The _Pulse of the City_.pdf.

[2] PulseCast project. OKF_Vault/mission/data.md, mission/mvp.md, mission/architecture.md, 2026.

[3] Code for Heilbronn. “Passantenzahlen messen mit LORA,” 18 September 2025. https://codeforheilbronn.de/blog/2025/09/passantenzahlen-messen-mit-lora/

[4] Code for Heilbronn. lorax / sensor-pax firmware and backend, 2026. https://codeberg.org/cfhn/lorax; upstream ESP32-Paxcounter and libpax, https://github.com/cyberman54/ESP32-Paxcounter and https://github.com/dbinfrago/libpax.

[5] PulseCast data-quality analysis, paper/research/07-data-quality.md, 2026.

[6] A. Kurkcu and K. Ozbay. “Estimating pedestrian densities, wait times, and flows with Wi-Fi and Bluetooth sensors.” Transportation Research Record 2644(1):72–82, 2017. DOI: 10.3141/2644-09.

[7] J. Martin, T. Mayberry, C. Donahue, L. Foppe, L. Brown, C. Riggins, E. C. Rye, and D. Brown. “A study of MAC address randomization in mobile devices and when it fails.” Proceedings on Privacy Enhancing Technologies 2017(4):365–383. DOI: 10.1515/popets-2017-0054.

[8] J. H. Friedman. “Greedy function approximation: a gradient boosting machine.” Annals of Statistics 29(5):1189–1232, 2001. DOI: 10.1214/aos/1013203451.

[9] L. F. Miranda-Moreno and A. C. Lahti. “Temporal trends and the effect of weather on pedestrian volumes: a case study of Montreal, Canada.” Transportation Research Part D: Transport and Environment 22:54–59, 2013. DOI: 10.1016/j.trd.2013.02.008.

[10] Bluetooth SIG. Bluetooth Core Specification v5.3, 2021. https://www.bluetooth.com/specifications/specs/core-specification-5-3/

[11] Court of Justice of the European Union. Case C-582/14, Patrick Breyer v Bundesrepublik Deutschland, 19 October 2016. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62014CJ0582

[12] European Data Protection Board. Guidelines 02/2023 on the Technical Scope of Article 5(3) of the ePrivacy Directive, 2023. https://www.edpb.europa.eu/documents/guideline/guidelines-22023-on-technical-scope-of-art-53-of-eprivacy-directive_en