Joseph M. Ernst

Influence of Vertical Sensor Placement on Data Collection Efficiency from Bluetooth MAC Address Collection Devices

The consumer electronics industry has made extensive use of the Bluetooth wireless protocol in many portable devices. A substantial number of these Bluetooth devices broadcast a unique identifier in the form of the media access control (MAC) addresses. These MAC addresses can be captured electronically and the same matching algorithms used in traditional license plate studies can be used to estimate segment travel time and origin-destination matrices. This paper briefly illustrates how these data can be used to estimate arterial link travel times and empirically illustrates the sensitivity of sample size to sensor placement. A controlled experiment with fixed lateral mounting and varying vertical mounting heights is then conducted to develop design recommendations for mounting Bluetooth monitoring devices. The paper concludes by recommending a Class I Bluetooth detector mounting height of at least 8 ft above the pavement grade. Based on a 24-h empirical data set on I-65 in Indianapolis, we found that 7.4%...

Performance Measures for Adaptive Signal Control: Case Study of System-in-the-Loop Simulation

The simulation of local signal controllers has become increasingly sophisticated in recent years and has been paralleled by improvements in the integration of adaptive systems into simulation. This paper describes and demonstrates an emerging methodology for the evaluation of adaptive signal control that is termed “system-in-the-loop simulation.” This methodology extends existing software-in-the-loop simulation by linking virtualized traffic controllers with real-world adaptive-control systems. In addition, the authors propose an analysis methodology that fuses data on simulated probe vehicles with data on high-resolution controller events. Through this data fusion, traditional measures of simulation performance such as delay can be enhanced with operational measures of performance that characterize quality of progression and capacity utilization. In addition, adaptive-control performance can be characterized in relation to overall impact on traveler delay and also described in terms that are meaningful for improvement of control schemes. An example case study is presented: the ACS-Lite adaptive system was tested on a 19-intersection system in Morgantown, West Virginia, under a special-event scenario. Free, fully actuated control was compared with traditional time-of-day and traffic-responsive control both with and without the use of the adaptive-control system ACS-Lite. Overall delay results are presented and contrasted with more detailed analysis of event-based performance measures at a single intersection and on a networkwide basis.

Non-linear compensation of vehicle signatures captured from electromagnetic sensors with application to vehicle re-identification

The distribution of travel times over a link in a transportation network can be estimated by observing the actual travel times of individual vehicles traversing the link and creating a histogram of the observations. The fundamental building block of travel time estimation is the re-idenfication of vehicles. A variety of techniques can be used for re-identification, but this paper focuses on electromagnetic signatures. The travel time distribution can be sampled using time-stamped signatures captured from vehicles along with an algorithm for signature matching. The method works well for free flowing traffic on a limited access highway, but to apply it on arterials it is necessary to compensate for vehicle acceleration over the sensors, which is manifested as a nonlinear scaling of the time axis in the captured signatures. In this paper we show how to estimate the acceleration from signatures captured from a loop trap and how to compensate for the effect of both varying velocities and acceleration. The results are summarized in terms of receiver operating characteristics to demonstrate the performance improvements associated with velocity and acceleration compensation. When applied to signature matching over a relatively long link it is shown that acceleration compensation offers significant improvement in matching performance.

Probe Data Sampling Guidelines for Characterizing Arterial Travel Time

Probe data are emerging as an important source for characterizing transportation systems. Travel time distributions have traditionally been characterized by the mean and standard deviation. These statistics work well to characterize uncongested freeway systems, which have travel time distributions that are approximately normal. When congested conditions or interrupted-flow facilities are encountered, the travel time distributions become more complex. Recently some additional travel time reliability indexes have been developed to quantify these travel time distribution characteristics. This study develops mathematical techniques for determining the sample size required for estimating the underlying travel time distributions that can be used for assessing changes in travel time distributions associated with operational changes of traffic signal controller offsets. The example provided shows that while gross changes in offsets require approximately 7 probe vehicle samples per study interval, subtle changes in offsets require approximately 80 probe vehicle data samples per study interval. Although these guidelines were developed for evaluating offset changes, the mathematical framework can be applied for evaluating the impact of other parameters, such as split times and cycle lengths. Further research on applying these mathematical techniques to a broader cross section of traffic conditions is warranted to assess their transferability to oversaturated conditions and freeways.

Maximum-likelihood speed estimation using vehicle-induced magnetic signatures

Modern traffic management systems require accurate vehicle detection, speed estimates, and link travel times for congestion detection, traveler information, ramp metering, optimization of traffic signal timing, and planning. Current speed estimation methods report speeds that are averaged over at least 30 seconds. This is necessary in some cases because the estimates tend to be noisy or in other cases because the algorithms are not intended to deliver individual vehicle speeds. This paper develops an algorithm based on communication theory and compares the results to conventional algorithms. The maximum-likelihood algorithm proposed in this paper provides significantly improved speed estimates that can be used to produce histograms of vehicle speeds instead of the speed averages currently available.

Recommended Tolerances for Magnetometer Orientation and Field Calibration Procedure

Magnetometers are increasingly being deployed for traffic detection in both fixed wired and wireless installations around the United States. Magnetometers respond to local magnetic field disturbances from the ambient magnetic field of the earth, and this response may be expected to depend on the relative orientation of the magnetometer axis and the direction of the earths magnetic field lines at the point of deployment. As documented by U.S. Geological Survey (USGS) maps, the declination and inclination of the earths magnetic field vary significantly over the continental United States, yet there is no traffic engineering design literature characterizing the effect of geographic magnetic field variation on magnetometer detection performance. This study briefly reviews the scientific theory, identifies the magnetometer installation parameters of most importance, and characterizes their impact on performance. In addition, recommendations are based on USGS data, which can be used to develop specifications for installation, testing, and acceptance of magnetometer-based detection technology. Finally, a simple procedure for field calibration of sensor orientation in a conduit to near-optimal rotation angle is presented.

Estimating Required Probe Vehicle Re-identification Requirements for Characterizing Link Travel Times

In the 1970s a framework was developed by Oppenlander to determine the sample size required for travel time estimation studies. This framework is still recommended today. This paper develops a new framework to improve upon the ideas set forth by Oppenlander. This new framework is based upon the Kullback-Leibler divergence. It allows for travel time studies to be evaluated in a more comprehensive way. Travel time estimation methods can now be evaluated on their ability to estimate travel time distributions instead of only the mean travel time. Also, this framework can be used on any travel time distribution whereas the Oppenlander framework was only properly suited for Gaussian distributions. The Kullback-Leibler divergence also allows for comparing both ID matching (i.e., small sample sizes with no erroneous travel times) and signature matching (i.e., large sample sizes mixed with some erroneous travel times) travel time estimation algorithms to be evaluated, while the Oppenlander framework was best suited for the ID matching algorithms. In this paper the Kullback-Leibler comparison framework for travel time studies is developed. The framework is then used to provide a comparison of an example ID matching and an example signature matching algorithm to demonstrate how both can be evaluated in a single framework. Finally, conclusions are made about the usefulness of the Kullback-Leibler comparison framework.

Crosstalk Detection in Signalized-Intersection Loop Detectors

Electromagnetic sensors, such as inductive loops and microloops, are widely used for vehicle detection. However, electrical coupling can occur between detection devices and result in electrical noise called crosstalk, which significantly degrades detection operation. Techniques such as channel scanning and frequency separation have been used to mitigate crosstalk; however, crosstalk is often transient because of environmental conditions and not always observed easily. This paper presents a crosstalk detection algorithm based on 156 h of training data taken from 12 sensors (inductive and microloop) and applies it to approximately 14 h of data collected from seven sensors (four inductive loops and three microloops) to demonstrate an automated method to identify periods when crosstalk occurs. The paper concludes by recommending the use of this approach for system acceptance and ongoing monitoring of sensor health.

Kullback-Leibler comparison framework for the evaluation of travel time distribution estimates

In the 1970s a framework was developed by Oppenlander for evaluation of travel time estimation studies. This framework is still recommended today. This paper develops a new framework to improve upon the ideas set forth by Oppenlander. This new framework is based upon the Kullback-Leibler divergence. It allows for travel time studies to be evaluated in a more comprehensive way. Travel time estimation methods can now be evaluated on their ability to estimate travel time distributions instead of only the mean travel time. Also, this framework can be used on any travel time distribution whereas the Oppenlander framework was only properly suited for Gaussian distributions. The Kullback-Leibler divergence also allows for both ID matching and signature matching travel time estimation algorithms to be evaluated, while the Oppenlander framework was best suited for the ID matching algorithms. In this paper the Kullback-Leibler comparison framework for travel time studies is developed. The framework is then used to provide a comparison of an example ID matching and an example signature matching algorithm to demonstrate how both can be evaluated in a single framework. Finally, conclusions are made about the usefulness of the Kullback-Leibler comparison framework.

Acceleration Estimation and Signature Matching Travel Time Estimation Enhancement for Standard Electromagnetic Vehicle Detectors

The traffic industry has invested in an extensive infrastructure for vehicle detection and speed estimation. This paper describes how these sensors can be enhanced to measure not only the presence and speed of a vehicle but also its acceleration and travel time. While acceleration of vehicles is of independent interest, knowledge of the acceleration of a vehicle can also be used to normalize vehicular signatures to increase significantly the probability of successfully reidentifying vehicles to generate travel time estimates. In this paper, the acceleration correction for vehicle reidentification is developed and analyzed in terms of the percentage of vehicles correctly matched and in the framework of a communication systems receiver operating characteristic. The percentage of vehicles correctly matched for the travel time estimation is increased from 55.1% when only the speed of the passing vehicles is used to 64.5% and the acceleration estimates are also used.