Traffic Data Collection via Cell Phone Signals

Traffic monitoring is nothing new. The passive collection of data using cameras, radar and other devices has been going on for years.

Mapping services from Microsoft  (Bing Maps), Yahoo! (Yahoo Maps) and Google (Google Maps) even provide convenient overlays for the state of traffic atop their maps.

Oh. Bing Maps and Google Maps also shows traffic on secondary roads, giving you a true opportunity to select a different route.

I posted “Tracking Movement and Progress via Bluetooth” back in May, citing the intent of the Indiana DOT to track vehicles and pedestrians using Bluetooth transmissions. Even though Bluetooth is a very close-range signal (less than 10 meters in most low-power situations), sensors could be placed to collect information from passing vehicles (or people).

This kind of crowdsourcing is very powerful; big globs of data collected by ambient means (sensors listening for signals) can be presented to a system ready to aggregate and report status in real time.

Note that I didn’t say “analyze”. Real-time analysis is left to the user (you need to look at the roads you might travel and decide which routes to avoid). Ideal use of mobile device, I’d say.

Analysts and engineers could rehydrate and create models with the collected data after the fact. Performing BI and data mining operations, they could advance some hypotheses to improve traffic.

Long story short: the better the data (quality and quantity), the higher the likelihood analysts the opportunity to evaluate current patterns and applying this knowledge, make improvements to the traffic systems.

• To improve quality, we want reasonable accuracy (over time) of a signal’s direction and location. From this, we can calculate average speed over distance. Knowing this speed will help analysts decide if there are enough lanes based on volume and average speed.
• Applying ambient and environmental data will also improve data quality: the time of day, weather conditions and events / activity at the traffic endpoints will help analysts identify anomalies that may affect overall averages.
• To improve quantity, we want more signals. We can improve this with more sensors and with collecting multiple types of signals. Rather than limiting collection to Bluetooth signals, add GPS and Cell signals to the dataset.

Comparing improved quality and quantity metrics to existing data collection methodologies will improve the ability of analysts and engineers to design better solutions and speed us on our way.