Conference PaperPDF Available

An Android Smartphone Application for Collecting, Sharing and Predicting Border Crossing Wait Time

Authors:

Abstract and Figures

This paper introduces an Android smartphone application called the Toronto Buffalo Border Wait Time (TBBW) app, designed to collect, share and predict waiting time at the three Niagara Frontier border crossings, namely the Lewiston-Queenston Bridge, the Rainbow Bridge, and the Peace Bridge. The innovative app offers the user three types of waiting time estimates: (1) current waiting times collected at the crossings; (2) historical waiting times; and (3) future waiting time predicted for the next 15 minutes and updated every five minutes. For the current waiting time, the app can provide both the data collected by border crossing authorities as well as user-reported or “crowd-sourcing” data shared by the community of the app’s users. Reporting of the data could be done either manually or automatically through a GPS tracking function provided by the smartphone. For the historical waiting time, the app provides statistical charts and tables to help users choose the crossing with the likely shortest wait time. Future waiting times are predicted by a real-time stepwise traffic delay prediction model which consists of a short-term traffic volume forecasting model and a multi-server queueing model. To validate the prediction functionality of the app, its predictions were compared against real-world delay measurements for the entire month of May, 2014. The comparison showed that the model offered predictions with a mean absolute difference of 9.22 minutes. When considering only delays that are greater than 10 minutes, the model has a mean absolute difference of only 6.95 minutes. The ability to integrate officially reported delay estimates with crowd-sourcing data, and the ability to provide future border wait times clearly distinguish the TBBW app from others on the market.
Framework of the stepwise delay prediction model 10 11 Border Crossing Traffic Volume Prediction Module: Three short-term traffic volume 12 prediction methods have been previously tested by the authors on the border crossing traffic 13 volume data for the Peace Bridge, namely seasonal Autoregressive Integrated Moving Average 14 (SARIMA), support vector regression (SVR), and an enhanced spinning network (SPN) (10-13). 15 In this app, SARIMA is chosen as the prediction method because of its easiness of 16 implementation and its moderate computational cost. As previously reported by the authors, for 17 a testing dataset with 1,905 hourly traffic volume points, the mean absolute percentage error 18 (MAPE) was found to be equal to 16.38% (10). It needs to be noted here that the short-term 19 traffic volume prediction module were built using data collected from the Peace Bridge, due to 20 the fine temporal resolution available (i.e., on the hourly basis) (5). The traffic volumes for the 21 other bridges were only available to the study on a daily basis at the time (6), and were thus 22 deemed not sufficient for accurate waiting time prediction. 23 24 Transient Multi-server Queueing Module: In the authors' previous work, 700 observations of 25 vehicular inter-arrival times and 571 observations of the service times (i.e. inspection time) were 26 collected from December 19, 2011 to January 10, 2012 at the Peace Bridge. Based on the 27 collected observations, it was determined that the distribution of the inter-arrival times is best 28 captured by an exponential distribution and that the service time distribution is best described as 29 an Erlang distribution with order equal to 2 (14). With these findings, an í µí±€/í µí°¸íµí°¸í µí±˜=2 /í µí±› queueing 30 model was developed to capture the queueing process at the border crossing. The transient 31 solution of this multi-server queueing model was then derived and used to predict the border 32 crossing waiting time (14). 33
… 
Content may be subject to copyright.
1
An Android Smartphone Application for Collecting, Sharing and Predicting
2
Border Crossing Wait Time
3
4
Lei Lin
5
Graduate Research Assistant
6
Department of Civil, Structural, and Environmental Engineering
7
University at Buffalo, the State University of New York, Buffalo, NY 14260
8
Phone: (716) 645-4347 FAX: (716) 645-3733
9
E-mail: llin22@buffalo.edu
10
11
Qian Wang, Ph.D.
12
Assistant Professor
13
Department of Civil, Structural, and Environmental Engineering
14
Institute for Sustainable Transportation and Logistics
15
University at Buffalo, the State University of New York, Buffalo, NY 14260
16
Phone: (716) 645-4365 FAX: (716) 645-3733
17
E-mail: qw6@buffalo.edu
18
19
20
Adel W. Sadek, Ph.D.***
21
Professor, Department of Civil, Structural, and Environmental Engineering
22
Director, Institute for Sustainable Transportation and Logistics
23
Director, Transportation Informatics University Transportation Center
24
University at Buffalo, the State University of New York, Buffalo, NY 14260
25
Phone: (716) 645-4367 FAX: (716) 645-3733
26
E-mail: asadek@buffalo.edu
27
28
29
Gregory Kott, Ph.D.
30
Research Scientist
31
PARC, A Xerox Company
32
E-mail: Gregory.Kott@xerox.com
33
34
Transportation Research Board
35
94th Annual Meeting
36
Washington, D.C.
37
*** Corresponding Author
38
Submission Date: August 1, 2014
39
Word Count: 4,909 text words + 8 Figures + 2 Tables = 7,409 equivalent words
40
41
42
Lin, Wang & Sadek 2
ABSTRACT
1
This paper introduces an Android smartphone application called the Toronto Buffalo Border
2
Wait Time (TBBW) app, designed to collect, share and predict waiting time at the three Niagara
3
Frontier border crossings, namely the Lewiston-Queenston Bridge, the Rainbow Bridge, and the
4
Peace Bridge. The innovative app offers the user three types of waiting time estimates: (1)
5
current waiting times collected at the crossings; (2) historical waiting times; and (3) future
6
waiting time predicted for the next 15 minutes and updated every five minutes. For the current
7
waiting time, the app can provide both the data collected by border crossing authorities as well as
8
user-reported or “crowd-sourcing” data shared by the community of the app’s users. Reporting of
9
the data could be done either manually or automatically through a GPS tracking function
10
provided by the smartphone. For the historical waiting time, the app provides statistical charts
11
and tables to help users choose the crossing with the likely shortest wait time. Future waiting
12
times are predicted by a real-time stepwise traffic delay prediction model which consists of a
13
short-term traffic volume forecasting model and a multi-server queueing model. To validate the
14
prediction functionality of the app, its predictions were compared against real-world delay
15
measurements for the entire month of May, 2014. The comparison showed that the model
16
offered predictions with a mean absolute difference of 9.22 minutes. When considering only
17
delays that are greater than 10 minutes, the model has a mean absolute difference of only 6.95
18
minutes. The ability to integrate officially reported delay estimates with crowd-sourcing data,
19
and the ability to provide future border wait times clearly distinguish the TBBW app from others
20
on the market.
21
22
Key Words: Border Crossing; Waiting Time; Crowd Sourcing; Short Term Traffic Volume
23
Prediction Model; Queueing Model; Android; Smartphone Application.
24
25
26
27
28
29
30
31
32
33
34
Lin, Wang & Sadek 3
1
INTRODUCTION
2
Due to the continuous travel demand increase, coupled with tighter security and inspection
3
procedures after September 11, border crossing delay has become a critical problem. As reported
4
by the Ontario Chamber of Commerce, border crossing delay causes an annual loss of
5
approximately $268.45 million for New York State. For the whole U.S., the cost is much higher
6
(1). According to a press release in 2008 given by the then U.S. Transportation Secretary, Mary
7
E. Peters, border delays cost Canadian and US businesses as many as 14 billion dollars in 2007
8
(2). Besides their negative economic impacts, border delays, and the associated idling of traffic
9
awaiting inspection, have a significant environmental cost. A 10-year study by Lwebuga Mukasa
10
et al. (3) showed a positive relationship between increased commercial traffic volume at Peace
11
Bridge border crossing between downtown Buffalo, New York and Fort Erie, Ontario, and the
12
increased use of asthma health care.
13
To address these issues, transportation authorities have recently begun to provide
14
travelers with information about current border crossing delays. This is the case for example in
15
the Buffalo-Niagara region, for example, where the Niagara International Transportation
16
Technology Coalition (NITTEC) has been providing such information to the public for years. In
17
the early years, the waiting time was obtained based on very rough and approximate estimates of
18
queue length. More recently, NITTEC is using blue-tooth identification technology to provide
19
more accurate delay estimates to motorists, and the information is now updated every five
20
minutes.
21
Regardless of the method however, there is an inherent limitation associated with
22
providing just the current border delay, which is likely to be quite different from the future wait
23
time that the travelers would experience by the time they arrive at the border. This is especially
24
true if there is a significant lag between the time when travelers need to act on the information
25
provided and the time of their arrival at the border. If the future waiting time can be predicted, it
26
would be more informative for travelers and businesses to select the time to depart and the route
27
to pursue. Moreover, with predicted border crossing delays, intelligent routing algorithms could
28
be developed to optimally direct and route border-destined traffic in a fashion that would
29
minimize the overall system travel time or the negative impacts on the environment.
30
The extremely data-rich environment of today provides an excellent opportunity for data
31
mining and for extracting useful insights to help improve transportation systems efficiency.
32
One more factor that deserves consideration is the emergence of social media applications using
33
smartphones which allow people to easily create, share and exchange information. For example,
34
Waze is a community-based traffic and navigation app, acquired by Google in 2013, where
35
drivers can share real-time traffic and road information, saving travel time, gas and money on
36
their daily commute (4).
37
This paper introduces an Android smartphone application (app) called the Toronto
38
Buffalo Border Wait Time (TBBW), which is designed to share waiting time among travelers of
39
the three Niagara Frontier border crossings, namely the Lewiston-Queenston Bridge, the
40
Rainbow Bridge, and the Peace Bridge. Three types of waiting times are offered based on users’
41
preferences, including the current waiting time, the historical waiting time, and the future waiting
42
time predicted by a real-time traffic delay prediction model.
43
Lin, Wang & Sadek 4
For the current waiting time, the app can provide both the data collected by the border
1
crossing authorities and the user-reported or “crowd-sourcing” data shared by the community of
2
users of the app. For the historical waiting time, the app provides statistical charts and tables to
3
help users choose the crossing with the likely shortest waiting time. Moreover, the app can also
4
provide future border waiting time for the next 15 minutes with an updating frequency of five
5
minutes. The future waiting times are predicted by a stepwise delay prediction model that
6
consists of a short-term traffic volume prediction model for predicting the incoming traffic flow
7
and a queueing model for predicting border inspection resulted delays.
8
9
PURPOSE AND SCOPE
10
The Niagara Frontier International Border includes three main bridges connecting Western New
11
York, U.S. to Southern Ontario, Canada, namely the Lewiston-Queenston Bridge, the Rainbow
12
Bridge, and the Peace. Figure 1 shows the yearly traffic volume going through Peace Bridge (one
13
of the three crossings) from 2009 to 2013. As can be seen, for each direction to U.S. or to
14
Canada, there are more than two million passenger vehicles and around 500,000 commercial
15
vehicles going through Peace Bridge every year. This highlights the very large market of
16
potential users and the great potential effect of this app. Besides that, thanks to the predictive
17
capabilities of TBBW, it can help border crossing and customs agencies determine the optimal
18
staffing level and the number of inspection booths needed to keep the border delay below a
19
certain threshold.
20
21
Lin, Wang & Sadek 5
1
FIGURE 1a Yearly traffic volume through Peace Bridge to U.S.
2
3
FIGURE 1b Yearly traffic volume through Peace Bridge to Canada
4
FIGURE 1 Yearly traffic volume of Peace Bridge from 2009 to 2013
5
The TBBW app was designed to collect, share and estimate border crossing waiting time
6
by taking advantage of multiple data sources and advanced traffic prediction methods. FIGURE
7
2 summarizes the characteristics of TBBW (shown in the green color), in comparison with the
8
existing border crossing delay dissemination method (shown in the blue color). As can be seen,
9
TBBW provides several options for border crossing delay estimates including, user-reported or
10
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
2009 2010 2011 2012 2013
Traffic Volume (vehicles)
Year
Passenger Vehicles Commercial Vehicles
0
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
2009 2010 2011 2012 2013
Traffic Volume (vehicles)
Year
Passenger Vehicles Commercial Vehicles
Lin, Wang & Sadek 6
“crowd-sourcing” wait time, historical, and future wait time, in addition to the current waiting
1
time reported by the authorities. Travelers and border management authorities can then make
2
better decisions based on this information.
3
4
FIGURE 2 Comparison of TBBW with the Other Ways to Share Border Waiting Time
5
6
DATASETS
7
Two types of data are used to develop the TBBW app. The first dataset contains the hourly
8
traffic volume data collected at the Peace Bridge since 2003. This is used as the input to develop
9
the stepwise border delay prediction model and to predict the future waiting times. The second
10
dataset captures the current waiting times collected and maintained by the border crossing
11
authorities. Such data are used as one source of the current waiting times provided by the app. In
12
addition, they are stored for historical data analysis and also used as the ground truth to assess
13
the performance of the border delay prediction model. All data are available for download from
14
Border Crossing
Travelers
Current waiting time
(provided by the bridge
authorities or users)
Historical waiting time
(documented by TBBW)
Future waiting time
(predicted by the stepwise
border crossing delay model)
For individuals: save time,
gas and money
For the system: improve
economic competitiveness
and minimize environmental
impacts
Current waiting time
(Official method)
Website
Telephone
Variable Message Board
Radio
Lin, Wang & Sadek 7
the websites maintained by the Peace Bridge authority and Niagara Falls Bridge Commission (5,
1
6).
2
INNOVATIVE FEATURES
3
The TBBW app was developed on the Android platform, the most popular mobile
4
operating system used in the U.S. As of July 2013, there are more than one million apps
5
available for Android in the Google Play Store (7). TBBW is available from the Google Play
6
store. As of 11/05/2014, there have been a total of 358 downloads. In terms of quality of reviews,
7
the app has received six “five stars” reviews and one “three stars” review. One comment is
8
available and which reads: “good work very responsive!”. The developed TBBW app is
9
innovative in terms of its ability: (1) to share current waiting time; (2) to store and analyze
10
historical waiting time; and (3) to predict future waiting time, as described below.
11
12
Sharing Current Waiting Time Function
13
14
15
FIGURE 3a Official Website FIGURE 3b Manually Share FIGURE 3c Automatic Share (GPS)
FIGURE 3 Three ways to share current waiting time
The app employs two ways to collect current waiting time information. The first way involves
downloading the waiting time data from the websites maintained by the Buffalo and Fort Erie
Public Bridge Authority and the Niagara Falls Bridge Commission. The current waiting time for
Peace Bridge and Lewiston Queen Bridge are provided and updated every five minutes (5), and
for Rainbow Bridge, it is updated every one hour (6). The information is collected and uploaded
in real time to the app as shown in FIGURE 3a.
Because the official current waiting time data is lagged (particularly for the Rainbow
Bridge where it is only updated every hour), the app also provides a second way to collect the
current waiting time data utilizing crowd sourcing ideas. Specifically, users are allowed to report
their experienced border crossing delays that can be then processed and broadcasted to other
users for their benefits (called crowd sourcing (8)). The same concept has been widely applied in
Lin, Wang & Sadek 8
other traffic information sharing apps, such as the pre-mentioned Waze and the bus arrival time
sharing app Tiramisu (9). In TBBW, users can share their waiting times by manually inputting
the data as shown in FIGURE 3b. They can also choose to automatically share their waiting
times through their GPS-enabled smartphones as shown in FIGURE 3c. This option is necessary
because if users are driving, it is unsafe and illegal to manually input waiting time.
Utilizing Historical Waiting Time Function
FIGURE 4a Average Waiting
Times for Each Day of Week
for Each Bridge
FIGURE 4b Comparison of
Waiting Times at Three Bridges
for the Past Hour
FIGURE 4c Waiting Times Sharing
History by the registered user
himself/herself
FIGURE 4 Three ways to utilize historical waiting time
1
Mining and analyzing historical border crossing waiting time data in a proper manner can
2
provide additional insight to travelers. In TBBW, three types of graphs and charts are created
3
based on an underlying historical waiting time database.
4
As can be seen in FIGURE 4a, for each bridge, the average waiting times for each day of
5
week are calculated and shown in one chart. This is the long term trend based on the historical
6
data of the past month. The TBBW app also allows the users to compare the waiting times of the
7
three bridges, based on the historical data of the past one hour, as shown in FIGURE 4b. Finally,
8
because users may want to make decisions based on their own previous experiences, registered
9
users can view their waiting times as another reference as shown in FIGURE 4c.
10
11
Predicting Future Waiting Time Function
12
Finally, in addition to current and historical analyses of wait times, the app is designed to predict
13
the likely waiting time in the next 15 minutes (this estimate is also updated every 5 minutes).
14
Predicting is based on utilizing the stepwise border crossing delay prediction model previously
15
developed by the authors (10-14). The following section will describe this model and its
16
prediction performance in detail.
17
Lin, Wang & Sadek 9
Stepwise Delay Prediction Model
1
The stepwise border delay prediction model is composed of two sequential modules as shown in
2
FIGURE 5 below. The first module is designed to predict the traffic volume arriving at the
3
border crossings for each time period (10-13). Note that the economic indicators and weather and
4
incident shown information in Figure 5 were not used in the current version of the short term
5
traffic volume prediction model; we hope to address this in our future research). Given the
6
predicted traffic volume as input, the second model estimates the corresponding waiting time by
7
solving a transient multi-server queueing problem (14).
8
9
FIGURE 5 Framework of the stepwise delay prediction model
10
11
Border Crossing Traffic Volume Prediction Module: Three short-term traffic volume
12
prediction methods have been previously tested by the authors on the border crossing traffic
13
volume data for the Peace Bridge, namely seasonal Autoregressive Integrated Moving Average
14
(SARIMA), support vector regression (SVR), and an enhanced spinning network (SPN) (10-13).
15
In this app, SARIMA is chosen as the prediction method because of its easiness of
16
implementation and its moderate computational cost. As previously reported by the authors, for
17
a testing dataset with 1,905 hourly traffic volume points, the mean absolute percentage error
18
(MAPE) was found to be equal to 16.38% (10). It needs to be noted here that the short-term
19
traffic volume prediction module were built using data collected from the Peace Bridge, due to
20
the fine temporal resolution available (i.e., on the hourly basis) (5). The traffic volumes for the
21
other bridges were only available to the study on a daily basis at the time (6), and were thus
22
deemed not sufficient for accurate waiting time prediction.
23
24
Transient Multi-server Queueing Module: In the authors’ previous work, 700 observations of
25
vehicular inter-arrival times and 571 observations of the service times (i.e. inspection time) were
26
collected from December 19, 2011 to January 10, 2012 at the Peace Bridge. Based on the
27
collected observations, it was determined that the distribution of the inter-arrival times is best
28
captured by an exponential distribution and that the service time distribution is best described as
29
an Erlang distribution with order equal to 2 (14). With these findings, an  queueing
30
model was developed to capture the queueing process at the border crossing. The transient
31
solution of this multi-server queueing model was then derived and used to predict the border
32
crossing waiting time (14).
33
Lin, Wang & Sadek 10
Because the TBBW app requires that the predicted wait time be updated every five
1
minutes, the predicted hourly traffic volume was split into a finer resolution (e.g., a five-minute
2
resolution) before they were used for border wait time prediction by the queueing models. With
3
the inter-arrival distribution known, this was done using the inverse cumulative function of the
4
inter-arrival exponential distribution    , where is the predicted hourly volume
5
or arrival rate.
6
Other input requirements of the queueing model included the number of inspection
7
booths. However, the number of open inspection stations is typically not available ahead of time.
8
To solve the problem, our approach at the moment involves running the queueing model for
9
different numbers of open stations (1 to 10 in this study), and trying to estimate how many
10
stations are actually open. Other venues to be explored in the near future are information offered
11
by users or directly by the U.S. Customs and Border Protection. The readers can find more
12
detailed information about these queueing models in the reference (14).
13
14
Prediction Results: The TBBW interface of the predicted waiting time for passenger vehicles
15
from Canada to U.S. through the Peace Bridge is shown in FIGURE 6.
16
17
FIGURE 6 Predicted border crossing waiting time
18
In order to test the prediction performance of the stepwise delay prediction model, the research
19
compared the predicted waiting times with the historical waiting times recorded by the border
20
authorities from 7:00 AM to 9:00 PM for each day of the whole month of May, 2014. Because
21
the future waiting time is updated every five minutes, there should be a total of 5,580 predicted
22
values for the month. However, because of several missing data points from the field
23
observations (e.g., when the server was down and the official waiting time was recorded as
24
“N/A”), a total of 3,103 observations were deemed valid for assessing the prediction model’s
25
performance.
26
27
TABLE 1 Prediction Performance of the Stepwise Delay Prediction Model
28
Lin, Wang & Sadek 11
Data Group
Number of data points
Mean Absolute Difference
(minutes)
Whole Dataset
3,103
9.22
Officially Recorded Waiting
Time = 0 minutes (denoting less
than 10 minute delays)
2,363
9.94
Officially Recorded Waiting
Time >=10 minutes
740
6.95
1
The mean absolute difference (minutes) between the predicted waiting times and the
2
officially recorded waiting times is shown in TABLE 1. As can be seen, the mean absolute
3
difference for the whole dataset is 9.22 minutes. After checking the officially recorded waiting
4
times, we find that there were a total of 2,363 data points where the wait times was recorded as
5
being equal to 0 minutes, and the remaining 740 points had delays greater than or equal to 10
6
minutes. After discussions with the border crossing authorities, it was revealed that their practice
7
was to report any wait time which was less than 10 minutes as 0 minutes delay. Given this, and
8
in order to provide for a true evaluation of the predictive model accuracy, the testing dataset was
9
split into two groups. The first group (2,363 data points) had an official reported delay of 0
10
minutes, which meant that the delay could be anywhere between 0 and 10 minutes. For that
11
group, the mean absolute difference between the model’s predictions and the officially reported
12
delay times was as high as 9.94 minutes (it should be clear now that that absolute error is
13
exaggerated, since the actual delay could have been anywhere between 0 and 10 minutes). The
14
second group included points where the officially reported wait time was greater than or equal to
15
10 minutes. For that second group, the mean absolute difference was only 6.95 minutes.
16
For a more disaggregate view of the performance of the delay prediction model, the
17
predicted waiting times and the historical waiting times for the peak hours 18:00-20:00 on April
18
22, 2014 are compared and shown on FIGURE 7.
19
20
Lin, Wang & Sadek 12
FIGURE 7 Prediction Performance for the peak hours of 18:00-20:00 on April 22, 2014
1
As can be seen in FIGURE 7, the mean absolute difference between the predicted waiting
2
times and the observations is about 6.6 minutes. Most of the time, the difference is within 10
3
minutes, except for 19:40 for which the difference is around 20 minutes. This is most probably
4
the result of the opening of additional inspection stations at that time without the model being
5
aware of that (the reader may recall that there is currently no easy way for the app to discern the
6
actual number of inspection stations open; it is hoped that in the future such information may be
7
obtained from the Customs and Border Protection agencies). Another reason could be that the
8
historical waiting time detected by the Bluetooth technology is lagging in time, since the
9
Bluetooth technology provides an estimate of the delay at the time a vehicle had joined the queue
10
some time prior to the reporting time (that time is actually equal to the time it took the vehicle to
11
exit the system).
12
Front-End Service Processes of Toronto Buffalo Border Wait Time (TBBW) app
13
Figure 8 shows the details of the TBBW front-end service processes behind the innovative
14
functions described above.
15
16
17
FIGURE 8 Flow Chart of TBBW Front-End Service Processes
18
19
As can be seen, there is a local computer which continuously runs the web crawler
20
program to download the current waiting times from the official border crossing authority
21
websites. That computer also continuously runs the step-wise border crossing waiting time for
22
24 hours per day. The current and predicted waiting times are then uploaded to the remote
23
database which is hosted by GoDaddy, an internet domain registrar and web hosting company
24
(15). Unlike the local computer, this remote server can be guaranteed to be running all the time,
25
Remote Database
(GoDaddy), 24/7
Border Crossing Authority Websites
web crawler to get the
current waiting times
Run the step-wise border
crossing waiting time
prediction model. 24/7
upload the current and
predicted waiting times
Upload the shared
waiting times
Download the current,
historical and
predicted waiting
times.
Store all the current,
historical and future
waiting times
Lin, Wang & Sadek 13
which is important for the app users, to allow them to interact with the server at any time. The
1
app users can upload their own experienced waiting times to the remote server, and can also
2
download different kinds of waiting times from it. The historical graphs and charts are generated
3
at the client side (i.e., the android smart phone).
4
5
COMPARISON WITH OTHER BORDER CROSSING APPS
6
A detailed comparison was conducted to demonstrate the innovative features of the TBBW app.
7
A few observations can be made based on TABLE 2 below. First, although all other border
8
crossing apps provide the waiting time for all border entries from Mexico to U.S. and from
9
Canada to U.S., none of them provide the waiting time for the travelers leaving the U.S. through
10
those borders. This is most probably because those apps all depend upon the data downloaded
11
from the same U.S. Customs and Border Protection website that only provides the waiting time
12
for travelers coming into the U.S. (16). In contrast, TBBW provides the waiting time for both
13
directions. It also would be a simple extension to expand TBBW to include all the Canada-US
14
and Mexico-US borders from that website.
15
16
TABLE 2 Comparison of TBBW with Other Border Crossing Apps
17
App name
Border Location
Bi-
direction
Update
Interval
Waiting Time Provided or
Not
Login
System
Crowd
Sourcing
From
To
Current
Historical
Future
Manual
GPS
TBBW
(17)
Toronto
Buffalo
Yes
5 min
Yes
Yes
Yes
Yes
Yes
Yes
Buffalo
Toronto
Best Time
to Cross
Border (18)
Mexico
US
No
5 min
Yes
Yes
No
No
Yes
No
Canada
US
Border
Wait Times
(19)
Mexico
US
No
N/A
Yes
No
No
No
No
No
Canada
US
Border
Times (20)
Mexico
US
No
N/A
Yes
No
No
No
No
No
Canada
US
Border Info
(21)
Mexico
US
No
N/A
Yes
No
No
No
No
No
Canada
US
Border
Check (22)
Mexico
US
No
N/A
Yes
No
No
No
No
No
Canada
US
18
Second, all the apps provide the current waiting time downloaded from the official
19
authorities, but only the “Best Time to Cross Border” app and our own app provide “crowd
20
sourcing based waiting time. The difference between TBBW and the “Best Time to Cross
21
Border” app, however, is that TBBW also provides one more option, that is, the ability to share
22
the waiting time automatically based on GPS location tracking, in addition to the ability to
23
manually input waiting time data. The fact that TBBW can automatically calculate and report
24
wait time is a huge advantage over the need for users to manually input the wait time themselves.
25
Third, only the Best Time to Cross Border” and TBBW utilize historical data. The
26
Best Time to Cross Border” app can produce a seven day comparison graph that compares the
27
Lin, Wang & Sadek 14
average waiting time for each day of the week, and the individual day graph that shows the
1
maximum, minimum and average waiting time at each hour for each day of the week. In
2
addition to those functionalities provided by the “Best Time to Cross Border” app, TBBW offers
3
an additional feature of comparing the waiting times at the three bridges for the past hour, and
4
can also show the sharing history for registered users.
5
Fourth, TBBW is the only app that has registration and login functionalities. While
6
unregistered users can still enjoy all the functions of this app, the ability to share waiting time
7
with others is restricted to registered users. The registration and login function is deemed useful
8
for a number of reasons. First, it can decrease the risk that some people share inaccurate waiting
9
time intentionally. Second, through registration, users can receive the latest notification about
10
border crossing traffic conditions. Lastly, as mentioned previously, the sharing history of
11
registered users can be recorded in the database, which can help them make decisions based on
12
their own experiences.
13
Finally, TBBW is the only one that can provide future waiting times. This is perhaps the
14
most significant advantage of TBBW, in contrast to other border crossing apps that simply
15
download and list the waiting times to the users. As shown in the previous section, the stepwise
16
border crossing delay prediction model, which is the engine behind this function, has been tested
17
and has a promising performance.
18
19
RISKS AND CHANLLENGES
20
This section will summarize the risks and challenges encountered while developing the app.
21
Some of those challenges have been addressed, while others are left for future work.
22
23
The Need for More Data
24
A critical piece of information for wait time prediction which is missing at this point is the
25
number of open lanes or inspection booths. Although the delay prediction model can estimate
26
the number of open lanes, it would be better and more accurate if the real value were to be
27
provided by the U.S. Customs and Border Protection agencies. The authors plan to work with
28
multiple agencies to explore methods for acquiring such data in the future.
29
Crowd Sourcing
30
As with any contribution-based crowd sourcing information system, a risk exists of low
31
motivation to participate and of abuse (23). To overcome this problem for TBBW, one can
32
design a set of reward and penalty rules on the basis of the registration and login function. For
33
example, when users share their border crossing waiting time with others, they can get some
34
virtual points, and every period of time the user with the highest rank may be rewarded. Abuse
35
can also be prevented through penalties. For example, users who intentionally share wrong
36
border crossing waiting times can be identified and filtered by setting a threshold for the
37
difference between the value provided by the user and a “best” estimate based on a combination
38
of the officially reported waiting time and the average waiting time from other users. Users who
39
abuse the system may also be restricted from sharing information.
40
GPS Location
41
Some privacy concerns may arise regarding the ability to share waiting time in an automatic
42
fashion through the GPS location sharing function. To address this, the TBBW app was
43
Lin, Wang & Sadek 15
designed so that it does not store any of the users’ GPS locations data; these data are only used to
1
calculate the distances of the travelers from the borders and their speed, so an approximate
2
waiting time can be estimated.
3
CONCLUSIONS AND FUTURE WORK
4
This paper introduced an android app TBBW which combines sophisticated transportation
5
models with emerging mobile computing technologies to solve the wait time border crossing
6
problem. The performance of the prediction model was assessed by comparing its predictions to
7
those reported by the authorities for month of May, 2014. The comparison demonstrated that the
8
predictions are quite accurate, with a mean absolute difference of only 6. 95 minutes for delays
9
greater than or equal to 10 minutes.
10
Several future directions are suggested by the current work. First, at the moment, the
11
TBBW app is only predicting the delay for the next 15 minutes, it would be better to make the
12
prediction horizon a user-specified value. Second, although the app is currently designed for the
13
Niagara International Frontier Borders, it can also be easily extended and applied to other US-
14
Canadian or US-Mexico borders. The app can even be extended to predict airport delay, and
15
delay at many other similar queueing systems, in the future.
16
17
ACKNOWLEDGEMENTS
18
Partial funding of this research has been provided by the Transportation Informatics Tier I
19
University Transportation Center headed at the University at Buffalo. The authors would like to
20
thank the University Transportation Center program for their financial support.
21
Lin, Wang & Sadek 16
REFERENCES
1
1. Ontario Chamber of Commerce (OCC). Cost of Border Delays to the United States Economy,
2
2005. Available online at:
3
http://www.thetbwg.org/downloads/Cost%20of%20Border%20Delays%20to%20the%20United
4
%20States%20Economy%20%20April%202005.pdf. Accessed on Oct 18, 2013.
5
2. U.S. Department of Transportation (USDOT), Office of Public Affairs. US Department of
6
Transportation Unveils New Program to Fight Border Congestion, 2008. Available at:
7
http://www.dot.gov/affairs/fhwa1208.htm. Accessed on Oct 18, 2013.
8
3. Lwebuga-Mukasa, J.S., S.J. Ayirookuzhi and A.Hyland. Traffic Volumes and Respiratory
9
Health Care Utilization among Residents in Close Proximity to the Peace Bridge before and after
10
September 11, 2001. Journal of Asthma, Vol. 40, No. 8, pp. 855-864, 2002.
11
4. Waze. Available at https://www.waze.com/. Accessed on 04/06/2014.
12
5. Peace Bridge. Available at www.peacebridge.com. Accessed on 04/12/2014.
13
6. Niagara Falls Bridge Commission. Available at http://niagarafallsbridges.com. Accessed on
14
04/12/2014.
15
7. Android. Available at http://en.wikipedia.org/wiki/Android_(operating_system)#Applications.
16
Accessed on 04/02/2014.
17
8. Crowd Sourcing. Available at http://www.merriam-webster.com/dictionary/crowdsourcing.
18
Accessed on 04/20/2014.
19
9. Zimmerman, J., A. Tomasic, C. Garrod, D. Yoo, C. Hiruncharoenvate, R. Aziz, and A.
20
Steinfeld. Field Trial of Tiramisu: Crowd-Sourcing Bus Arrival Times to Spur Co-design. In
21
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1677-
22
1686, 2011.
23
10. Lin, L., Q. Wang, S. Huang and A. W. Sadek. On-line Prediction of Border Crossing Traffic
24
using an Enhanced Spinning Network Method. Transportation Research Part C: Emerging
25
Technologies, 2013.
26
11. Lin, L., A. W. Sadek and Q. Wang. Multiple-Model Combined Forecasting Method for
27
Online Prediction of Border Crossing Traffic at Peace Bridge. In Transportation Research Board
28
91st Annual Meeting (No. 12-3398), 2012.
29
12. Lin, L., Q. Wang and A. W. Sadek. Evaluating Short-term Traffic Volume Forecasting
30
Models based on Multiple Datasets and Data Diagnosis Measures. In Transportation Research
31
Board 92nd Annual Meeting (No. 13-3691), 2013.
32
13. Lin, L., Q. Wang and A. W. Sadek. Short-Term Forecasting of Traffic Volume Evaluating
33
Models Based on Multiple Data Sets and Data Diagnosis Measures. Transportation Research
34
Record: Journal of the Transportation Research Board, Vol. 2392, No. 1, pp. 40-47, 2013.
35
Lin, Wang & Sadek 17
14. Lin, L., Q. Wang and A. W. Sadek. Border Crossing Delay Prediction Using Transient Multi-
1
server Queueing Models. Transportation Research Part A: Policy and Practice, Vol. 64, pp. 65-
2
91, 2014.
3
15. GoDaddy. Available at https://www.godaddy.com. Accessed on 11/05/2014.
4
16. U.S. Customs and Border Protection. Available at
5
http://bwt.cbp.gov/?com=1&pas=1&ped=1&plist=0901. Accessed on 03/06/2014.
6
17. Toronto Buffalo Border Waiting. Available at
7
https://play.google.com/store/apps/details?id=toronto.buffalo.borderwaiting&hl=en. Accessed on
8
11/05/2014.
9
18. Best Time to Cross The Border. Available at
10
https://play.google.com/store/apps/details?id=net.calit2.ports&hl=en. Accessed on 11/05/2014.
11
19. Border Wait Times. Available at
12
https://play.google.com/store/apps/details?id=com.sopinka.bwt&hl=en. Accessed on 11/05/2014.
13
20. BorderTimes. Available at
14
https://play.google.com/store/apps/details?id=com.geogrant.bordertimes&hl=en. Accessed on
15
11/05/2014.
16
21. Border Info. Available at
17
https://play.google.com/store/apps/details?id=com.jflores.dev.borderinfo&hl=en. Accessed on
18
11/05/2014.
19
22. BorderCheck. Available at
20
https://play.google.com/store/apps/details?id=com.ssamuel.bordercheck&hl=en. Accessed on
21
11/05/2014.
22
23. Steinfeld, A., J. Zimmerman, A. Tomasic, D. Yoo, and R. D. Aziz. Mobile Transit
23
Information from Universal Design and Crowdsourcing. Transportation Research Record:
24
Journal of the Transportation Research Board, Vol. 2217, No. 1, pp. 95-102, 2011.
25
26
... The authors of this paper have been studying border crossing delay problems for more than two years. We have brought the time series concepts in estimating the time-of-day travel pattern (11); proposed a two-step delay flow prediction model that consists of a short-term traffic volume prediction model and a queueing model (11,12); In 2015, we even released an Android Smartphone Application for Collecting, Sharing, and Predicting Border Crossing Waiting Time (13). Besides these studies, seldom has analyzed the bi-national delay patterns for both commercial and passenger vehicles through the three bridges at Niagara Frontier border. ...
... • If not, what factors may lead to the uneven distribution of border delays over the three bridges? Our yearly efforts have collected enough resources to deliver our research purposes: first, our developed Android Application can filter out useful information from the official website of the Niagara Frontier border crossing authorities (13) and collect the border delay data on these three bridges shown in Figure 1; second, another java applications are built to collect the weather data from weather website (14) and can be included in the influential factor analysis; besides, the time-of-day, season of the year as well as the special days are also labeled for the study. ...
Article
Full-text available
Border crossing delays between New York State and Southern Ontario cause problems like enormous economic loss and massive environmental pollutions. In this area, there are three border-crossing ports: Peace Bridge (PB), Rainbow Bridge (RB) and Lewiston-Queenston Bridge (LQ) at Niagara Frontier border. The goals of this paper are to figure out whether the distributions of bi-national wait times for commercial and passenger vehicles are evenly distributed among the three ports and uncover the hidden significant influential factors that result in the possible insufficient utilization. The historical border wait time data from 7:00 to 21:00 between 08/22/2016 and 06/20/2017 are archived, as well as the corresponding temporal and weather data. For each vehicle type towards each direction, a Decision Tree is built to identify the various border delay patterns over the three bridges. We find that for the passenger vehicles to the USA, the convenient connections between the Canada freeways with USA I-190 by LQ and PB may cause these two bridges more congested than RB, especially when it is a holiday in Canada. For the passenger vehicles in the other bound, RB is much more congested than LQ and PB in some cases, and the visitors to Niagara Falls in the USA in summer may be a reason. For the commercial trucks to the USA, the various delay patterns show PB is always more congested than LQ. Hour interval and weekend are the most significant factors appearing in all the four Decision Trees. These Decision Trees can help the authorities to make specific routing suggestions when the corresponding conditions are satisfied.
... Second, the authors of this paper have long studied the problem of predicting border crossing delay at the Peace Bridge (Lin et al., 2012(Lin et al., , 2013b(Lin et al., , 2014a(Lin et al., , 2014b(Lin et al., , 2015b. Our previous work has clearly demonstrated that traffic patterns at the border are impacted significantly by factors like holidays, sport events, weather and day of the week. ...
Article
In this paper, we aim to quantify uncertainty in short-term traffic volume prediction by enhancing a hybrid machine learning model based on Particle Swarm Optimization (PSO) and Extreme Learning Machine (ELM) neural network. Different from the previous studies, the PSO-ELM models require no statistical inference nor distribution assumption of the model parameters, but rather focus on generating the prediction intervals (PIs) that can minimize a multi-objective function which considers two criteria, reliability and interval sharpness. The improved PSO-ELM models are developed for an hourly border crossing traffic dataset and compared to: (1) the original PSO-ELMs; (2) two state of the art models proposed by Zhang et al. (2014) and Guo et al. (2014) separately; and (3) the traditional ARMA and Kalman filter models. The results show that the improved PSO-ELM can always keep the mean PI length the lowest, and guarantee that the PI coverage probability is higher than the corresponding PI nominal confidence, regardless of the confidence level assumed. The study also probes the reasons that led to a few points being not covered by the PIs of PSO-ELMs. Finally, the study proposes a comprehensive optimization framework to make staffing plans for border crossing authority based on bounds of PIs and point predictions. The results show that for holidays, the staffing plans based on PI upper bounds generated much lower total system costs, and that those plans derived from PI upper bounds of the improved PSO-ELM models, are capable of producing the lowest average waiting times at the border. For a weekday or a typical Monday, the workforce plans based on point predictions from Zhang et al. (2014) and Guo et al. (2014) models generated the smallest system costs with low border crossing delays. Moreover, for both holiday and normal Monday scenarios , if the border crossing authority lacked the required staff to implement the plans based on PI upper bounds or point predictions, the staffing plans based on PI lower bounds from the improved PSO-ELMs performed the best, with an acceptable level of service and total system costs close to the point prediction plans.
... The authors of this paper have long studied the problem of predicting border crossing delay at the Peace Bridge (22)(23)(24)(25)(26). Our previous work has clearly demonstrated that the traditional, single-value prediction approach cannot capture the dynamics of border crossing traffic volumes very well (23)(24). ...
Article
Full-text available
Short-term traffic volume prediction models have been extensively studied in the past few decades. However, most of the previous studies only focus on single-value prediction. Considering the uncertain and chaotic nature of the transportation system, an accurate and reliable prediction interval with upper and lower bounds may be better than a single point value for transportation management. In this paper, we introduce a neural network model called Extreme Learning Machine (ELM) for interval prediction of short-term traffic volume and improve it with the heuristic particle swarm optimization algorithm (PSO). The hybrid PSO-ELM model can generate the prediction intervals under different confidence levels and guarantee the quality by minimizing a multi-objective function which considers two criteria reliability and interval sharpness. The PSO-ELM models are built based on an hourly traffic dataset and compared with ARMA and Kalman Filter models. The results show that ARMA models are the worst for all confidence levels, and the PSO-ELM models are comparable with Kalman Filter from the aspects of reliability and narrowness of the intervals, although the parameters of PSO-ELM are fixed once the training is done while Kalman Filter is updated in an online approach. Additionally, only the PSO-ELMs are able to produce intervals with coverage probabilities higher than or equal to the confidence levels. For the points outside of the prediction levels given by PSO-ELMs, they lie very close to the bounds.
... Previously, the authors of this paper have proposed a two-step delay prediction model that consists of a short-term traffic volume prediction model for predicting the incoming traffic flow and a queueing model for predicting border inspection resulted delays (3)(4)(5)(6)(7). An Android smartphone application called the Toronto Buffalo Border Wait Time (TBBW) app is also developed to collect, share and predict waiting time at the three Niagara Frontier border crossings (8). ...
Article
Full-text available
Border crossing delays cause problems like huge economics loss and heavy environmental pollutions. To understand more about the nature of border crossing delay, this study applies a dictionary-based compression algorithm to process the historical Niagara Frontier border wait times data. It can identify the abnormal spatial-temporal patterns for both passenger vehicles and trucks at three bridges connecting US and Canada. Furthermore, it provides a quantitate anomaly score to rank the wait times patterns across the three bridges for each vehicle type and each direction. By analyzing the top three most abnormal patterns, we find that there are at least two factors contributing the anomaly of the patterns. The weekends and holidays may cause unusual heave congestions at the three bridges at the same time, and the freight transportation demand may be uneven from Canada to the USA at Peace Bridge and Lewiston-Queenston Bridge, which may lead to a high anomaly score. By calculating the frequency of the top 5% abnormal patterns by hour of the day, the results show that for cars from the USA to Canada, the frequency of abnormal waiting time patterns is the highest during noon while for trucks in the same direction, it is the highest during the afternoon peak hours. For Canada to US direction, the frequency of abnormal border wait time patterns for both cars and trucks reaches to the peak during the afternoon. The analysis of abnormal spatial-temporal wait times patterns is promising to improve the border crossing management
Article
The improvement of 128 bit architectures has analyzed IPv6, and current trends suggest that the improvement of sensor networks will soon emerge. Here, we prove the simulation of link-level acknowledgements, which embodies the significant principles of artificial intelligence. In our research, we investigate how congestion control can be applied to the study of online algorithms.
Article
Researchers agree that cacheable methodolo-gies are an interesting new topic in the field of partitioned theory, and mathematicians concur. Our purpose here is to set the record straight. Given the current status of interactive archetypes, system administrators particularly desire the emulation of extreme programming. We describe an analysis of ran-domized algorithms, which we call CAUTER.
Article
Unified robust epistemologies have led to many practical advances, including the lookaside buffer and write-back caches. In fact, few end-users would disagree with the construction of kernels [1]. Our focus in this position paper is not on whether simulated an-nealing and write-back caches are always incompatible , but rather on motivating a novel system for the exploration of semaphores (Trysail).
Article
Full-text available
Although several methods for short-term forecasting of traffic volume have recently been developed, the literature lacks studies that focus on how to choose the appropriate prediction method on the basis of the statistical characteristics of the data set. This study first diagnosed the predictability of four traffic volume data sets on the basis of various statistical measures, including (a) complexity analysis methods, such as the delay time and embedding dimension method and the approximate entropy method; (b) nonlinearity analysis methods, such as the time reversibility of surrogate data; and (c) long-range dependency analysis techniques, such as the Hurst exponent. After the data sets were diagnosed, three models for short-term prediction of traffic volume were applied: (a) seasonal autoregressive integrated moving average (SARIMA), (b) knearest neighbor (k-NN), and (c) support vector regression (SVR). The results from the statistical data diagnosis methods were then correlated to the performance results of the three prediction methods on the four data sets to determine the means for choosing the appropriate prediction method. The results revealed that SVR was more suitable for nonlinear data sets, while SARIMA and k-NN were more appropriate for linear data sets. The data diagnosis results were also used to devise a selection process for the parameters of the prediction models, such as the length of the training data set for SARIMA and SVR, the average number of nearest neighbors for k-NN, and the input vector length for k-NN and SVR.
Conference Paper
Full-text available
Crowd-sourcing social computing systems represent a new material for HCI designers. However, these systems are difficult to work with and to prototype, because they require a critical mass of participants to investigate social behavior. Service design is an emerging research area that focuses on how customers co-produce the services that they use, and thus it appears to be a great domain to apply this new material. To investigate this relationship, we developed Tiramisu, a transit information system where commuters share GPS traces and submit problem reports. Tiramisu processes incoming traces and generates real-time arrival time predictions for buses. We conducted a field trial with 28 participants. In this paper we report on the results and reflect on the use of field trials to evaluate crowd-sourcing prototypes and on how crowd sourcing can generate co-production between citizens and public services.
Article
Extensive interviews with riders of the Pittsburgh, Pennsylvania, bus system revealed that, as the top priority, riders wanted to know the actual arrival time of buses. Following a universal design approach, a system called Tiramisu was created to foster a greater sense of community between riders and transit bus service providers. The design focused on acquisition of crowdsourced information for bus location and bus fullness. On the basis of that input, the system predicted the arrival time of buses and provided a convenient platform for reporting problems and positive experiences within the transit system. The intention was to create a community of riders that materially participated in the delivery of the transit service. Tiramisu also supported specific information and reporting needs for riders with disabilities and thereby provided greater independent mobility around the community. An early field trial of Tiramisu suggested that the approach was both feasible and potentially viable.
Article
A recent study based on data over a 10-year period (1991-2000) showed a positive association between health care utilization and prevalence of asthma, and commercial traffic at a U.S.-Canada border crossing. We wanted to determine whether decreases in total traffic would also be associated with decreases in health care utilization for respiratory illnesses. Following September 11, 2001, there was a 50% drop in total traffic at the Peace Bridge border crossing point between Buffalo, New York and Fort Erie, Ontario, Canada. To investigate the impact of such a traffic decline on health care utilization for respiratory illnesses, weekly respiratory admissions to Kaleida Health System, Western New York's largest health care provider were analyzed according to ICD9CM classification and compared with total weekly traffic volumes for 3-month periods in 2000 and 2001 (August, September, and October). The total number of patients admitted to hospital or seen in emergency departments for respiratory illnesses during the 3-month periods of both years was 5288. A 50% drop in total traffic following Labor Day and September 11, 2001, from week 4 to week 7 was found to be statistically significant (p = 0.031) when a one-way ANOVA was performed. Likewise, the drop in total respiratory cases approached statistical significance (p = 0.052) when a one-way ANOVA was conducted. The results suggest an association between decrease in traffic volumes with decrease in health care utilization for respiratory diseases. These results suggest that current levels of traffic may be impacting on the respiratory health of residents in the nearby community.
Multiple-Model Combined Forecasting Method for 27
  • L Lin
  • A W Sadek
  • Q Wang
Lin, L., A. W. Sadek and Q. Wang. Multiple-Model Combined Forecasting Method for 27
Available at www.peacebridge.com. Accessed on 04
  • Peace Bridge
Peace Bridge. Available at www.peacebridge.com. Accessed on 04/12/2014. 13
Available at https://www.godaddy.com. Accessed on 11
  • Godaddy
GoDaddy. Available at https://www.godaddy.com. Accessed on 11/05/2014. 4