TABLE 1 List of Freeway Corridors Included in Data

Document Sample
TABLE 1 List of Freeway Corridors Included in Data Powered By Docstoc
					Comparative Analysis of Speed Measurements of Work Zone Speed Enforcement
Equipment and Ground-Based Traffic Data Station

Ching-Yao Chan, Ph.D., P.E.
California PATH, Headquarters
Institute of Transportation Studies
University of California, Berkeley
Richmond Field Station, Bldg. 452, 1357 S. 46th Street
Richmond, CA 94804, USA
TEL: 510-665-3621, FAX: 510-665-3757, cychan@path.berkeley.edu

Kang Li
California PATH, Headquarters
Institute of Transportation Studies
University of California, Berkeley
Richmond Field Station, Bldg. 452, 1357 S. 46th Street
Richmond, CA 94804, USA
TEL: 510- 665-3552, FAX: 510-665-3757, kangli@berkeley.edu


ABSTRACT

Speeding is a significant contributor to a significant portion of highway collisions, for work
zones in particular. Automated speed enforcement is one potential solution to reduce the number
of collisions in work zones where speeding is a persistent problem. This paper describes a recent
study that was undertaken to assess the technical performance of work-zone automated speed
enforcement (ASE) equipment in the field. Several commercially-off-the-shelf traffic monitoring
devices, along with a selective ASE system, RWASS, were field tested at selected study sites in
California. The study site was located on a rural two-lane highway, where speeding appeared to
be common. The RWASS tends to yield a lower measured value of speed measurement, with a
mean differential of less than 1 mph. The standard deviation of the speed differential is between
3 to 4 mph. The assessment of technical performance of ASE and other traffic monitoring
devices can offer insights in the process of validating functional characteristics and seeking
performance enhancements. The outcome of this study will provide valuable support for future
ASE implementation.




                                               1
1.0 INTRODUCTION

Speeding is a significant safety concern. For work zones in particular, the speeding problem is
compounded by on-site road re-configuration, narrowed lanes, or poor visibility. As part of the
efforts to seek safety improvements of the California highway network, the California
Department of Transportation (Caltrans) is exploring the implementation issues of automated
speed enforcement (ASE) and sponsoring a project conducted at California PATH 1 , University
of California at Berkeley.

An earlier phase of the project was carried out and concluded with an extensive literature review
and an examination of various institutional and legal issues involved in the implementation of
ASE [1]. The prior work was followed with a series of field experiments to validate the technical
performance of such ASE systems. A work-zone ASE system was acquired and installed in a
field experimental site, along with several other commercially-off-the-shelf traffic monitoring
devices. This paper provides a description of the field experiment and the outcome of data
analysis.

2.0 TECHNICAL EVALUATION OF ASE EQUIPMENT

With the advancements in sensing and communication technologies, a variety of traffic
monitoring devices are becoming more affordable and feasible. For the purpose of implementing
ASE as well as traffic monitoring devices, it is important for highway operators and enforcement
agencies to define and to understand the performances of such equipment. More importantly,
selective components and sub-systems can potentially be integrated to provide traffic
enforcement and management functions more economically and effectively. Therefore, one
major objective in the current study is to conduct a comparative evaluation of several candidate
traffic monitoring systems so that their field performance can be fairly and thoroughly
investigated. Preliminary results of traffic data gathered with several commercially-off-the-shelf
products have been documented and reported. [2-3] A brief summary of the equipment used in
the field experiments is given below.

2.1 Experimental Data Collection Plans

In a separate project, the research team has surveyed and tested a number of commercially-off-
the-shelf traffic monitoring devices for a different application [4-6]. Based on previous
experience of evaluating such products, a select set of candidate products were acquired and
developed for the current task.

Commercially-off-the-shelf traffic sensors were selected that could potentially offer comparative
output for the evaluation of ASE equipment. The combined sensor suite was deployed at selected
locations along the case-study highway. The data collection sites were chosen preferably to
allow field data collection with variations in setup configurations and traffic patterns. At each

1
    www.path.berkeley.edu



                                                2
location, the data collection process continued for approximately one week. During the course of
the field observation, data were collected for comparative evaluation but no enforcement
function was executed.

2.2 Caltrans Traffic Data Station

Caltrans has a traffic monitoring station near Post-Mile 5.6 of Route 12, close to the data
collection site. At this station, a pair of in-ground loops is spaced by 12 feet (3.6 meters) with a
piezoelectric sensor in between. The double loops and the piezoelectric sensor are combined to
provide vehicle count, speed measurement, vehicle length, and axle spacing for each passing
vehicle. The Caltrans Traffic Data Station is considered the most reliable data source because
the station was validated during installation and the results across multiple data show consistent
output. Given its combination of double loop construction, at least the vehicle count
measurement should be fairly accurate. The collected data from this station were used as the
baseline to compare to those measured by other sensing devices on the eastbound lane.

2.2.1 Innovations at Traffic Data Station (Courtesy of Clint Gregory of District 10, Caltrans)

The traffic data station was a joint effort between InfoTek Wireless 2 and Caltrans District 10.
They created a customized solution combining the InfoTek Wizard hardware, its Intelligent Loop
Detection Application (ILDA) firmware, and data collector middleware to solve its several
identified problems and to provide near real-time performance measurement data.

The new low-cost platform enabled District 10 to provide near real-time data directly for the
Freeway Performance Measurement System [7-8], developed in cooperation with the University
of California at Berkeley and other partners; it allowed for data sharing with Traffic Census, not
previously achievable; it facilitated the capture of much more detailed and extensive traffic data;
it made possible the archiving of traffic data; and it provided an upgrade to the Caltrans
Automated Warning System (CAWS). The project was undertaken as a long-term solution with
a projected life of at least ten years.

Caltrans District 10’s rollout of the InfoTek Wizard platform advances ITS by its use of a cutting
edge, bi-directional GPRS intelligent modem for wireless data collection and M2M
communication, its ILDA software and middleware for collecting, analyzing and archiving more
precise data (including actual speed, occupancy, and classification of individual vehicles on a
lane-by-lane basis), and its Web-based, user-intuitive GUI interface—all rolled out over existing
infrastructure.

By using existing infrastructure down to the cabinet level, significant cost savings were realized,
and the solution was able to be implemented very quickly, requiring minimal installation
procedures and, being a wireless solution, no laying of costly cable or complex wiring. Once the
new data was in place, it immediately began providing data that helped the Caltrans meet its

2
    http://www.infotekwireless.com/



                                                 3
PLAP (Program Level Action Plan) goals for improving safety and mobility for California’s
travelers.

With the InfoTek Wizard platform, Caltrans District 10 is achieving a level of performance not
possible with the previous system. First, the system allows for the capture of a much higher level
of detail in traffic data. Then, that traffic data is now archived and is being analyzed to improve
safety and mobility within the system. Finally, in addition to capturing traffic data, the system
monitors the health of the components. When a component fails, rather than sending a field
engineer to examine 15 miles of highway until he determines which site and which component is
faulty, the InfoTek Wizard platform detects the faulty component and almost instantly notifies
the specified Caltrans District 10 employees by email and pager. The engineer can often even
repair the failed component remotely. The result is significantly shorter downtime, and a reduced
number of and reduced duration for field service trips. The bottom line is exponential cost
savings, time savings and increased efficiency. The schematic diagram in Figure 1 depicts the
interaction of Infotek Wizard with other system components in an integrated system.




Figure 1 Schematic diagram InfoTek Wizard in an Integrated System (Courtesy of Mike
Poursartip of Infotek Wireless)




                                                4
In addition to the projected benefits of implementing this system, Caltrans District 10
experienced some unanticipated side benefits. The cutting-edge Java-based technology employed
by the InfoTek Wizard hardware means that more than 12 InfoTek Wizard units can be deployed
in the space formerly occupied by one 170 collector. This saves storage space, as well as time
and effort for installation and maintenance personnel. Further, Caltrans District 10 has achieved
substantive, measurable power savings of up to 50% by employing the InfoTek Wizard platform,
which uses only milliamps per unit, compared to the old hardware that used several amps per
unit.

2.3 Traffic Monitoring Devices

For the data collection task, an industrial computer (PC-104) system was developed and
interfaced with a traffic monitoring radar and an ASE equipment to record synchronized data.
Four different types of sensors shown in Figure 2 were jointly used in the filed experiments. For
more detailed descriptions, refer to earlier reports [2].

The sensor suite used in the filed included:
• Road Working Area Safety System (RWASS) manufactured by Sensys Traffic 3 AB, Sweden
• EVT-300 TM (Eaton-Vorad Technologies) 4 Tracking Radar
• Nu-Metrics NC-200 TM (Quixote Technology) 5 Traffic Sensors
• Trans-Q TM (Quixote Technology) 6 Radar Traffic Classifier




         RWASS                        EVT-300                     NC-200               Trans-Q

               Figure 2 Devices Used in Data Collection for Technical Evaluation

2.4 Equipment Configuration and Layout

A schematic diagram in Figure 3 depicts the arrangement of equipment.
• The Caltrans traffic data station is located 250 feet (75 meters) upstream of the RWASS
   location.
• RWASS and EVT-300 were set up on roadside with the radar antennas oriented to the
   upstream direction to cover oncoming traffic.
3
  http://www.sensys.se/
4
  http://www.roadranger.com/Roadranger/productssolutions/collisionwarningsystems/index.htm
5
  http://www.qttinc.com/pages/nc200.html
6
  http://www.qttinc.com/newproducts/trans-Q.html



                                                      5
•   A trailer equipped with solar panels and batteries was located further from traffic lanes to
    provide power supplies for RWASS and EVT-300.
•   Trans-Q was mounted on a pole attached to the trailer. This unit is oriented in a 45-degree
    direction opposite from the other radar. The trans-Q lateral position was located within 30
    feet from the centerline of the near traffic lane.
•   Four NC-200 devices were installed on the near lane using adhesive tapes. The four sensors
    were arranged sequentially to acquire data at different distances relative to the radar location.
•   A data acquisition computer was stored inside the trailer box. A synchronization signal is
    provided by RWASS to EVT-300 data computer whenever RWASS was triggered by a
    speeder traveling over the pre-determined threshold.




                                 Traffic

            RWASS
            EVT-300                        Traffic Data Station
            NC-200                                                                  Trailer
            Trans-Q

                 Figure 3 Layout of Equipment Setup at Data Collection Site

3.0 SUMMARY OF DATA COMPARISON AMONG DIFFERENT DEVICES

A preliminary account of traffic data collected from the Caltrans data station and other traffic
monitoring devices has been reported in an earlier publication [3]. This section first provides a
review of the results and the conclusions reached from field observations in July 2008 on State
Route 12. This roadway has a posted speed limit of 55 mph.

3.1 SR-12 Caltrans Traffic Station

Table 1 Caltrans Vehicle Count with Percentage of Vehicle Traveling at Speed of 65 mph
or Higher
Date                                07-16     07-17      07-18     07-19      07-20
Vehicle Count                         8499       8861      9732       8898       7127
Vehicle Counts of Speed > 65 mph        274       272        243       452        366
Percentage of Counts > 65 mph        3.22%     3.07%      2.50%     5.08%      5.14%
Night time (Hours 0-6)                                                      
Vehicle Count                           466       454        515       408        350
Vehicle Counts of Speed > 65 mph         57        46         57        69         62
Percentage of Counts > 65 mph       12.23% 10.13% 11.07% 16.91% 17.71%




                                                    6
The Caltrans traffic station was activated throughout the one-week study period, but was
intermittently turned off for maintenance reasons for the later part of the week. Table 1 shows
the vehicle counts and the percentage of speeders for 5 consecutive days. Note that in the last
two columns, for July 19 (Saturday) and July 20 (Sunday), the percentages of speeding vehicles
are much higher than the other days. In particular, the ratio of speeding vehicles is the lowest on
Friday, when the traffic was the heaviest. Additionally, the total and speeder counts are also
listed for the night time hours, from midnight to 6 o’clock in the morning. There is a significant
increase in the ratio of speeding vehicles during this period.

3.2 RWASS

RWASS was activated throughout the collection period of one week. However, only 4-plus days
of data were collected due to vandalism and loss of power supply. The triggering threshold for
RWASS was set at 65 mph, 10 mph higher than the speed limit, to take into account that the
traffic flow in this region generally moves at a speed higher than the speed limit. Out of 22,849
targets detected by RWASS, the system registered 671 times (2.9%) of speeding records for
vehicles traveling at 65 mph or higher.

Table 2 RWASS Vehicle Count with Percentage of Vehicle Traveling at Speed of 65 mph or
Higher
Date                              07-16     07-17      07-18     07-19     07-20*
Vehicle Count                       5373      5471       6124      5560       321*
Vehicle Counts of Speed > 65 mph      144       130        131       221       45*
Percentage of Counts > 65 mph      2.68%     2.38%      2.14%     3.97% 14.00%*
Night time (Hours 0-6)                                                    
Vehicle Count                         402       380        406       308       273
Vehicle Counts of Speed > 65 mph       32        27         38        41        38
Percentage of Counts > 65 mph      7.96%     7.11%      9.36% 13.31% 13.92%*

3.3 Summary of Data Analysis and Performance of Sensor Suite

Based on an overall review of data from various traffic monitoring systems, the following
observations were made:
(1) The Caltrans Traffic Data Station is considered the most reliable data source and used as the
    baseline for comparative analysis.
(2) The other surface-based sensing devices, NC-200, have vehicle counts most compatible with
    the Traffic Station.
(3) RWASS underestimates the count of passing vehicles due to its relatively low mounting
    position at approximately one meter and occlusion of some vehicles in traffic streams.
(4) Trans-Q has vehicle counts compatible with Traffic Station and NC-200, but it appears to
    have underestimated speed measurements.




                                                7
(5) Despite the differences in vehicle counts, several data sources provided very compatible
    estimates of speeder population in the range of 2-3 % on weekdays and 4-5% on weekends.
    The speeding threshold was set at 65 mph in a zone with a 55-mph speed limit.
(6) During the nighttime hours, the ratio of vehicles exceeding the speeding threshold increased
    several folds to the range of 7-10 % on weekdays and 14-17% on weekends.

4.0 DATA ASSOCAITION AND COMPARATIVE ANALYSIS OF SPEED
MEASUREMENTS

The techniques of signal processing and statistical analysis are applied to investigate the
performance of speed measurements provided by RWASS, with the Caltrans data station serving
as the baseline. There are mainly two tasks performed in order to analyze and compare the
RWASS and Caltrans loop sensor data. One is data synchronization and the other is data
association. Descriptions of the techniques and the results are given in the following sections

4.1 Synchronization and Association of Data Series

Since the time stamps of data sequence for RWASS and Caltrans loop sensors were set
separately in experiments, the data of the two systems need to be synchronized through post-
processing. After synchronization, data association between the two data sets can be performed
for further data analysis. An example of the synchronization result is shown in Figure 4.
                                                 RWASS vs Loop Sensor Data (Non-synchronized)
                                                                                                       Loop Sensor
                                                                                                       RWASS
                            70


                            60
              Speed (mph)




                            50
      (a)
                            40


                            30

                            10.6   10.7   10.8    10.9      11       11.1       11.2   11.3     11.4    11.5     11.6
                                                                  Time (hour)

                                                  RWASS vs Loop Sensor Data (Synchronized)             Loop Sensor
                                                                                                       RWASS
                            70


                            60
              Speed (mph)




                            50

      (b)
                            40


                            30

                            10.6   10.7   10.8    10.9      11       11.1       11.2   11.3     11.4    11.5     11.6
                                                                  Time (hour)


Figure 4 Synchronization of RWASS and Caltrans Loop Sensor Data. (a) Non-
synchronized (b) Synchronized



                                                                    8
Although the RWASS and loop sensor have quite different sensor characteristics in terms of
accuracy, sensitivity, robustness, and sampling rate, etc. due to the sensors’ intrinsic properties,
the two data sets should exhibit similar traffic pattern/dynamics. Based on this concept, cluster
analysis and pattern recognition were conducted to identify the time lag between the two data
sets. Figure 4 (a) and (b) show the speed versus time data before and after synchronization,
respectively. It can be observed that without synchronization, the two data series do not match in
time. On the other hand, the synchronized data sequences have corresponding data points that
line up reasonably well side by side.

Figure 5 shows an example of the data association for RWASS and loop sensor. Theoretically,
the missed detection rate of RWASS is higher than the Caltrans loop sensor due to missed targets
that were occluded by leading vehicles. Therefore, the total vehicles detected by RWASS can be
regarded as a subset of the total vehicles detected by Caltrans loop sensor. Also, each data value
represents a different vehicle. As a result, for each RWASS data value, there exists only one
corresponding value in the loop sensor data.

                                                                                                           RWASS
                                                Data Association for RWASS and Loop Sensor Data
                                                                                                           Loop Sensor
                                70
                                                                                                           Associated Loop Sensor Data
                  Speed (mph)




                                60

                                50
       (a)
                                40

                                10.62   10.64           10.66         10.68           10.7         10.72           10.74         10.76
                                                                          Time (hour)
                                                 Data Association for RWASS and Loop Sensor Data            RWASS
                                                                                                            Loop Sensor
                                70
                                                                                                            Unassociated RWASS Data
                  Speed (mph)




                                60

                                50

       (b)                      40

                                10.62   10.64           10.66         10.68           10.7         10.72           10.74
                                                                          Time (hour)
                                                Data Association for RWASS and Loop Sensor Data            RWASS
                                                                                                           Associated Loop Sensor Data
                                70
                  Speed (mph)




                                60

                                50

                                40
       (c)
                                10.62   10.64           10.66         10.68           10.7         10.72           10.74         10.76
                                                                          Time (hour)


             Figure 5 Data Association for RWASS and Caltrans Loop Sensor Data

After data synchronization, data association between RWASS and loop sensor data is carried out
based on the proximity in time. Namely, for each RWASS data point, the closest (in time) loop
sensor data point is chosen as the “associated” loop sensor data point representing the same
vehicle detected by RWASS and loop sensor. As shown in Figure 5 (a), the light blue circles
represent the loop sensor data that have been associated with the RWASS data. Figure 5 (b)



                                                                             9
shows the same data association result, but the light blue circles represent the RWASS data
which are not able to be associated with any loop sensor data. The lack of associated loop sensor
data for the RWASS data might be due to either the missed detection of the loop sensor or the
sensor error of the RWASS. Thus, these RWASS data are excluded in our data analysis. Figure 5
(c) shows all RWASS and associated loop sensor data. It can be seen that the occurrences of the
associated loop sensor data points and their corresponding RWASS data points match quite well
although there are some RWASS data not being able to be matched to any loop sensor data.

4.2 Speed Differential Distribution

More in-depth analysis of the data from RWASS and ground-loop data station was conducted.
The following figures use the sample set from July 19 as a case study to illsutrate the results
from the analysis.

Figure 6 provides an hour-by-hour count of data points for the two data series from RWASS and
Caltrans data station. Figure 6(a) depicts the difference in vehicle counts, and Figure 6(b) the
speed differentials for associated data pairs. It can be seen that the RWASS consistently lags in
vehicle counts, and it is most obvious during the day time when traffic is heavy.




           (a)




           (b)




Figure 6 Sensor Data Comparison (July 19, 2008). (a) Vehicle Detection Counts by RWASS
                        and Loop Sensor. (b) Speed Difference




                                               10
Once data association is carried out on the two sequences, the speed differential can be obtained
for each associated pair. The statistical mean and standard deviation (STD) for the data sets in
each one-hour period are then calculated. Figure 7 plots the mean and STD values for the 24-
hour period. The mean values for the two data sources are quite close, especially in the day-time
period. However, it can still be observed that the loop-based data have a slightly higher mean
values. The differential becomes more amplified in the midnight hours. The standard deviations
for the loop-based data also have a higher value than RWASS, which means that the range of
values is wider for the loop-based data station.

                                                  Mean of Speed
                               65
                                                                           RWASS
                                                                           Loop Sensor
                 Speed (mph)




                               60

          (a)
                               55


                               50
                                    0   5        10             15        20             25
                                                    Time (hour)
                                            Standard Deviation of Speed
                               10                                          RWASS
                                                                           Loop Sensor
                               8
                 STD (mph)




                               6
         (b)
                               4

                               2
                                    0   5        10                 15    20             25
                                                      Time (hour)

Figure 7 RWASS and Loop Sensor Speed Data Analysis. (a) Mean Value of Speed (b)
Standard Deviation of Speed

Figure 8 shows the relation between Speed Difference and Vehicle Class, where Speed
Difference is defined as (RWASS speed – Loop Sensor speed), and Vehicle Class is determined
based on vehicle length estimated by Caltrans data station. Vehicle length is shortest for Class 1
and longest for Class 4. Figure 8(a) and 8(b) shows the mean and STD of speed differential for
this data set. It can be seen that both the mean and STD decrease as the vehicle class/vehicle
length increases.

Figure 9 shows the relation between Speed Difference and the measured vehicle speed. It
appears that the discrepancy between the RWASS and Loop Sensor speed measurements is
larger at extreme low speeds. In other words, it is possible that the measurements of either
RWASS or loop-based stations become unstable at the very low-speed range.




                                                        11
                                                                            Mean Value of Speed Difference vs Vehicle Class
                                                  -0.2

                                                  -0.4



                               Mean Value(mph)
                                                  -0.6
          (a)
                                                  -0.8

                                                        -1

                                                  -1.2
                                                             1        1.5               2              2.5            3        3.5         4
                                                                                                  Vehicle Class
                                                                               STD of Speed Difference vs Vehicle Class
                                                        4


                                                       3.5
                                     STD (mph)




          (b)                                           3


                                                       2.5


                                                        2
                                                             1        1.5               2              2.5            3        3.5         4
                                                                                                  Vehicle Class


 Figure 8 Analysis of Speed Difference versus Vehicle Class (July 19, 2008). (a) Mean of
            Speed Difference (b) Standard Deviation of Speed Difference.

                                                                                   Mean of Speed Differential vs Speed
                                                       50
                               Mean (mph)




                                                        0
          (a)                                          -50

                                                 -100
                                                     10          20    30        40         50   60      70       80     90    100   110   120
                                                                                              Max Speed (mph)
                                                                                      STD of Speed Differential vs Speed
                                                       20
                                           STD (mph)




          (b)                                          10


                                                        0
                                                         10      20    30        40         50   60      70       80      90   100   110   120
                                                                                            Max Speed (mph)
                                                                                       Number of total vehicles: 22361
                                           15000
                Counts of Vehicles




                                           10000
          (c)
                                                 5000

                                                        0
                                                        10       20    30        40         50      60    70     80       90   100   110   120
                                                                                                 Max Speed (mph)



Figure 9 Analysis of Speed Difference versus Speed (July 16~19, 2008). (a) Mean of Speed
                 Difference (b) Standard Deviation of Speed Difference.




                                                                                                     12
Table 3 lists the parameters of data sets from four complete days of filed experiments. The range,
maximum, minimum, mean and standard deviations are given. It can be seen that the results of
analysis are consistent and stable across multiple data sets, which supports the validity of the
approaches taken in data processing.

Table 3 Data Comparison of RWASS and Loop-Based Data Sets

Date (July 2008)                                     16th           17th           18th           19th

Number of Total RWASS Data                                  5373           5471           6214           5560
Number of Associated RWASS Data                             5316            5436           6085           5524
Successful Association Rate (%)                             98.94          99.36          99.36          99.35

Maximum Loop Sensor Speed (mph)                             116.7           89.5           81.3           83.9
Minimum Loop Sensor Speed (mph)                               1.7            2.3           15.3              4
Mean Value of Speed from Loop Sensor                        55.61          55.72          54.41           56.2
STD of Speed from Loop Sensor                                5.95           5.81           5.96           5.81

Maximum RWASS Speed (mph)                                      82             81             80             89
Minimum RWASS Speed (mph)                                      12             10             11             20
Mean Value of Speed from RWASS                              54.84          55.02          53.78          56.18
STD of Speed from RWASS                                      6.60           5.41           5.90           5.46

Maximum Speed Difference (mph)                               51.1           60.7           40.2           65.5
Minimum Speed Difference (mph)                              -65.7          -33.9          -41.7          -31.6
Mean of Speed Difference (mph)                              -0.52          -0.53          -0.30          -0.38
STD of Speed Difference (mph)                                3.98           3.81           3.61           3.82

Mean Value of Associated Loop Sensor Speed
                                                            55.39          55.56          54.13          56.56
(mph)
STD of Associated Loop Sensor Speed (mph)                    6.60           6.22           6.52           6.34
Mean Value of Associated RWASS Speed (mph)                  54.87          55.02          53.83          56.17
STD of Associated RWASS Speed (mph)                          5.24           5.39           5.73           5.43

5.0 CONCLUDING REMARKS

This paper describes a recent study that was undertaken to assess the speed-measurement
performance of automated speed enforcement (ASE) equipment through field experimental



                                               13
design. Several commercially-off-the-shelf traffic monitoring devices, along with a selective
ASE system, RWASS, were field tested at selected study sites in California. The study site was
located on a rural two-lane highway, where speeding appeared to be common. Data collected
through the use of multiple traffic sensors provided compatible estimates of speeder population.

Data processing techniques, including synchronization and association, were adopted to
investigate in depth the speed measurement discrepancies between the selected ASE equipment
and a ground-loop based data station, which was taken as the baseline with the consideration of
its prior validation and calibration. The speed enforcement unit tends to yield a lower measured
value of speed measurement, with a mean differential of less than 1 mph. The standard deviation
of the speed differential is between 3 to 4 mph.

The results from the field experiments revealed that traffic speed measurements are likely to
yield discrepancies. Since speed measurement consistency and accuracy are major concerns in
the implementation of speed enforcement systems, it is critical that the operation of enforcement
take into account the characteristics of field performance of such devices and set speeding
thresholds accordingly. For a more robust and reliable system, it will also be desirable to utilize
technical approaches to offer supplementary speed measurements.

The assessment of technical performance of ASE and other traffic monitoring devices offers
significant insights in the process of validating functional characteristics and seeking
performance enhancements. The outcome of this study, in conjunction with the experience and
knowledge gained by other agencies in their development and implementation of work-zone and
general ASE systems, can provide valuable support for future ASE implementation.

                                  ACKONWLEDGEMENTS

The authors wish to express gratitude for assistance from sensor suppliers and subcontractors,
including Joe Jeffrey, Tony Espinoza, and Jack Carr, who have been extremely supportive during
the execution of the project. We are particularly grateful for the tremendous support provided by
Duper Tong, Joe Silvey and Clint Gregory of Caltrans District 10 and Asfand Siddiqui of
Caltrans Headquarters. We are indebted to the support of my colleagues, Thang Lian, Jeff Ko,
Bart Duncil, David Nelson, and Susan Dickey, who made the data collection possible.

This work was performed as part of a project (PATH Task Order 6212) sponsored by the
California PATH Program of the University of California, in cooperation with the State of
California Business, Transportation and Housing Agency, Department of Transportation. The
contents of this paper reflect the views of the author, who are responsible for the facts and
accuracy of the data presented herein. The contents do not necessarily reflect the official views
or policies of the State of California.

REFERENCES




                                                14
(1) C. J. Rodier, S. A. Shaheen, E. Cavanagh, “Automated Speed Enforcement for California: A
    Review of Legal and Institutional Issues,” California PATH Research Report, UCB-ITS-
    PRR-2007-14, September 2007.
(2) C-Y. Chan, “Field Evaluation of Work-Zone Automated Speed Enforcement and Traffic Monitoring
    Devices,” TRB 2009 Annual Meeting, Paper No. 09-1732, January 2009.
(3) C-Y. Chan, “Technical Evaluation of Road Working Area Safety Systems and Traffic sensors,
    “ Scientific Paper, Proceedings of the 2008 World Congress on Intelligent Transportation Systems,
    New York, New York, November 2008.
(4) A. Sharafsaleh, C-Y Chan, “Experimental Evaluation of Commercially-off-the-shelf Sensors
    for Intersection Decision Support Systems,” Technical Paper, Proceedings of the 2005
    Intelligent Transportation System World Congress in San Francisco, November 2005.
(5) C-Y. Chan, et al., “California Intersection Decision Support: A Systems Approach to
    Achieve Nationally Interoperable Solutions,” California PATH Research Report, UCB-ITS-
    PRR-2005-11, April 2005.
(6) J. A. Misener, et al., “California Intersection Decision Support: A Systems Approach to
    Achieve Nationally Interoperable Solutions II,” California PATH Research Report, UCB-
    ITS-PRR-2007-1, January 2007.
(7) Chen, C., Petty, K., Skabardonis, A., Varaiya, P., Jia, Z. Freeway Performance Measurement
    System: Mining Loop Detector Data, Annual Meeting of the Transportation Research Board,
    Washington, D.C., Jan 2001.
(8) Choe, T., Skabardonis A., and Varaiya. P. Freeway Performance Measurement system
    (PeMS): An operational analysis tool TRB 81st Annual Meeting, January 2002.




                                                 15