Ticon and traditional technologies. Comparative analysis of traffic data collection methods

Updated: Dec 24, 2018

Ticon and traffic detectors

Metric: Hourly Volume

Compared to: Miovision Video Detector

Hardware free vs Hardware-based


Direct comparison of Ticon estimation and Miovision measurements. Miovision measurements had been verified by manual counting from video recording.


The accuracy of Ticon algorithm using consolidated input data available from local transportation departments is about the same as traditional proven detection technology can deliver.

Ticon and traffic counters Metric: AADT

Compared to: Traditional traffic counters

Hardware free vs Hardware-based


Literature review and direct comparison of Counters measurements to Ticon estimations. Traffic counter used for the study: Georgia DOT, permanent stations [1]. Total number of estimations – 2074. Counters locations: city roads with low seasonal traffic fluctuations. For counters, each 48 h average had been compared with 365 days average. For Ticon, AADT estimation based on 365 days average had been compared with 365 day average from permanent station.


The use of ‘nearby’ traffic counts for AADT estimation even at 1 mi distance leads to the error of 30-150% [2]. Inaccuracy of larger spans is unpredictable. Relative error for AADT estimation based on usual 48 h counting may exceed 50% has MRE 12.3% ,which is about the same accuracy as estimated by counters on 1 week base (fig. above, see also Fig. 1-ApdxTicon estimation for AADT - green line). Even though Ticon method shows slight larger MRE with 90% probability, the risk of getting misleading results from Ticon is lower (maximal registered relative error is below 30%, while counters had shown 50% error, and, according to [2], it may exceed 75%. Ticon ensure 95% plus spatial coverage, which makes Ticon AADT incomparably more accurate than estimated by traditional methods.

Ticon and floating cars

Metric: Travel time reduction.

Compared to: Floating cars

95%+ time coverage vs 5%- time coverage

Ample results vs Bias and Sparse results


Total number of floating car runs – 76. Number of Ticon estimation - 192. Site: Fremont, CA [3]. Direct comparison of travel time measurements performed by floating cars with travel time estimations made by Ticon. Direct comparison of the ‘before-after’ benefits calculated from floating cars’ measurements and Ticon estimations.


Travel time value measured by floating cars are always in three-SIGMA range of vehicles speed distribution. Still, they shifted from true median (see Fig. 3-APDX and 4-APDX). MRE observed - about 19%. Since expected travel time reduction from ITS implementation usually in the range 10% - 30%, calculation of travel time reduction based on floating cars’ measurements can lead to bias results. Additionally, floating car technology presumes limited number of runs and its time coverage does not exceed 5% that adds to uncertainty of the results.


  1. Traffic Counts in Georgia

  2. ESTIMATES OF AADT: QUANTIFYING THE UNCERTAINTY. Shashank Gadda, et al., 86th Annual Meeting of the Transportation Research Board, Washington, D.C., January 2007.

  3. Jia Hao Wu, Gregory Brodski. Intelligent transportation system development and opportunities: Practical Cases for Smart Cities in China and US. 2017, SVCTBA Smart City Conf. Expo, San Jose, June 2, 2017


Ticon and traffic counters


  • The use of 48 h counting for AADT estimation may lead to unpredictable error, which can exceed 30% even for major streets with lo traffic seasonal fluctuations.

  • AADT based on one week counting can be expected of about 20% accuracy, but only month plus counting can guarantee better result

  • Above conclusions well supported by [2], where even larger errors originated from insufficient temporal coverage of counting had bees observed.

Fig. 1 - APDX, Metric: AADT

  • Portable and permanent detection has insufficient spatial coverage that leads to the necessity of spatial approximation with the use of ‘nearby’ detector.

  • The use of ‘nearby’ traffic counts for AADT or traffic volume estimation even at one mile distance leads to the error of 30-150% [2]. Inaccuracy of larger spans is unpredictable.

Fig. 2 - APDX, Metric: AADT and volume

Ticon and floating cars


  • Most of us instinctively think that all parts of the flow move with the same speed. That is not true for the traffic flow.

  • Vehicle speeds in traffic flow distributed normally as shown on the example on the graph.

Fig. 3 - APDX, Metric: Travel Time Distribution inside Traffic Flow

  • Ticon obtains direct measurements of travel time from 3% - 20% of traveling vehicles, which means at least 1000 veh/day for regular street, while floating cars technology performs 20 runs per day at maximum.

  • Ticon delivers statistically significant value of median travel time with 95 % reliability and MRE less than 5%. Floating cars deliver statistically insignificant results with the discrepancy of 200%+, while median offset is 19%

Fig. 4 - APDX, Metric: Travel Time Reduction

Recent Posts

See All

Impact of Autonomous Vehicles on City Traffic

Educated sources optimistically declare that autonomous vehicles (AV) could solve traffic problems by driving more efficiently and coordinating with other vehicles and infrastructure. While efficient