Sensor Placement for Urban Traffic Interpolation: A Data-Driven Evaluation to Inform Policy

TL;DR

This work tackles the challenge of citywide traffic-volume interpolation under sparse sensor coverage by benchmarking data-driven sensor-placement strategies across Berlin and Manhattan using real-world Strava and taxi data. It systematically compares spatial categories (centrality, feature-based, spatial dispersion, and active learning) and temporal designs, using $XGBoost$ as the interpolation model and metrics such as $MAE$ and $RMSE$. The main finding is that spatially even coverage (notably spatial dispersion) and simple, uncertainty-aware active learning substantially improve interpolation accuracy, with as few as $K=10$ sensors achieving large reductions in error; temporally, deploying single-day observations across many locations and even weekday distribution further enhances performance. Importantly, temporary deployments can closely approximate the performance of permanently deployed networks when scheduled optimally, offering cities substantial flexibility in budget and logistics. The study provides transferable, policy-relevant guidance for designing data-driven sensor networks that enhance reliability, equity, and cost-effectiveness in urban traffic monitoring.

Abstract

Data on citywide street-segment traffic volumes are essential for urban planning and sustainable mobility management. Yet such data are available only for a limited subset of streets due to the high costs of sensor deployment and maintenance. Traffic volumes on the remaining network are therefore interpolated based on existing sensor measurements. However, current sensor locations are often determined by administrative priorities rather than by data-driven optimization, leading to biased coverage and reduced estimation performance. This study provides a large-scale, real-world benchmarking of easily implementable, data-driven strategies for optimizing the placement of permanent and temporary traffic sensors, using segment-level data from Berlin (Strava bicycle counts) and Manhattan (taxi counts). It compares spatial placement strategies based on network centrality, spatial coverage, feature coverage, and active learning. In addition, the study examines temporal deployment schemes for temporary sensors. The findings highlight that spatial placement strategies that emphasize even spatial coverage and employ active learning achieve the lowest prediction errors. With only 10 sensors, they reduce the mean absolute error by over 60% in Berlin and 70% in Manhattan compared to alternatives. Temporal deployment choices further improve performance: distributing measurements evenly across weekdays reduces error by an additional 7% in Berlin and 21% in Manhattan. Together, these spatial and temporal principles allow temporary deployments to closely approximate the performance of optimally placed permanent deployments. From a policy perspective, the results indicate that cities can substantially improve data usefulness by adopting data-driven sensor placement strategies, while retaining flexibility in choosing between temporary and permanent deployments.

Figures (12)

• Figure 1: Street networks and sensor locations in the two study areas. The Berlin cycling network consists of 4,958 street segments and 34 bicycle counting stations (21 permanent and 13 temporary). The Manhattan street network comprises 8,156 street segments and 8 temporary traffic sensors. Dots mark sensor locations; in Manhattan, several sensors are clustered spatially, with four nearly overlapping in Midtown. For visualization purposes, street segments are colored by average traffic volumes over the respective observation periods, although the analysis uses daily Strava counts for Berlin and hourly taxi counts for Manhattan. Insets show the distributions of daily and hourly traffic volumes, respectively.
• Figure 2: Temporary placement considerations of temporary sensors: The upper row compares deployment strategies that vary the number of days allocated to each location, effectively trading off between broader spatial coverage (more locations, fewer days each) and repeated measurement at fewer locations, while keeping the total number of observations constant. The lower row examines whether restricting temporary measurements to specific weekdays affects prediction accuracy. Results are shown for Berlin (left) and Manhattan (right). The reported error metric is mae.
• Figure 3: mae performance under temporary and permanent sensor deployments for citywide traffic volume estimation, shown as a function of the sensor budget $K$.
• Figure 4: Distribution of per-fold mae values across sensor locations under logo cross-validation.
• Figure 5: Distributions of ground-truth sensor measurements for both cities. Berlin measurements are recorded at the daily level, whereas Manhattan measurements are recorded at the hourly level.