This paper reports on a survey utilizing Bluetooth detectors to quantitatively measure the operations of airdrome security checkpoints. Bluetooth detectors were deployed at George Bush Intercontinental Airport from 28 August to 13 September, 2010 and collected over 200,000 travel clip records between 9 Stationss located along entree roads, parking garages, non-sterile countries of terminuss, and unfertile countries of terminuss. In add-on, hourly checkpoint rider volumes were collected. The overpowering volume of this information necessitated the development of statistical word picture processs and artworks that characterize both the travel clip records and checkpoint information. These checkpoint operations studies provide dashboard flat studies utile for direction to quickly measure operations, and place chances for improved allotment of staffing. The paper concludes by supplying illustrations of how the checkpoint operations study constructs can be extended to qualify the motion of traffic through the airdrome, the full rider reaching experience, and numerically qualify user hold in person-hours.
Over the past decennary, security lines at airdromes have received considerable national attending as new showing processs have been introduced. Although the showing processs are new, they must be implemented in terminus installations that are in many instances decennaries old with important infinite restraints.
Figure 1 shows a exposure of three different airdrome security waiting lines. The waiting lines in both Figure 1a and Figure 1b are rather long. However, from both the system operator and user position, the more of import public presentation step is the length of clip it takes to pass through through the line which is one of the public presentation step mentioned by Correia et. Al with respects to measuring airport degree of service ( 9 ) . Although there are a assortment of factors that affect the clip to pass through through a security waiting line, the most important impact on a theodolite clip is the figure of lanes unfastened. Figure 1c shows an illustration of a security line in Houston with a modest waiting line and merely one of the two showing lanes unfastened. Manataki and Zografos have worked on a composite theoretical account for terminal analysis which includes security showing, but have merely evaluated this against package simulations ( 8 ) . Without quantitative delay clip informations throughout the twenty-four hours for that checkpoint and for alternate terminal checkpoints, it is really hard to measure the effectivity of checkpoint staffing degrees.
Obtaining both historical and real-time measurings of the clip it takes riders to pass through through both the waiting line and security showing procedure at airdrome checkpoints is of involvement to a assortment of stakeholders.A For illustration, riders would profit from this information by leting them to schedule appropriate slack clip in their reaching at the airport.A Managers responsible for staffing security checkpoints at a specific airdrome would hold quantitative information to agenda forces necessary to react to daily, seasonal, and particular event/holiday traffic. Zografos and Madas have proposed a determination devising procedure which leverages bing theoretical accounts and simulation package, but recognize that future demands and successful determination devising tools would necessitate include a user-friendly interface that does n’t necessitate extended preparation ( 10 ) . Finally, airdrome operators could utilize historical informations from checkpoints throughout their airdrome to ease long term planning for redevelopment and capital undertakings.
On a national graduated table, quantitative security testing theodolite clip informations for major airdromes across the state would let authorities functionaries to be more nimble in reacting to alterations in airdrome origin-destination traffic forms and air hose agendas. For illustration, in the Spring of 2009 it was estimated there were about 46000 full clip tantamount security screeners employed by the Transportation Security Administration. Assuming a to the full laden wage rate ( including benefits ) of $ 27.79/hour, this corresponds to an approximative one-year outgo of about $ 2.6 billion on testing forces. Assuming real-time quantitative rider theodolite clip informations could let directors to cut down staffing by 2 % , due to more effectual staff programming, this corresponds to an approximative one-year nest eggs of $ 53 million.
Extra benefits would besides accrue to the going public by directors utilizing quantitative tools to consistently cut down security checkpoint holds that occur due to switching travel patterns.A
Data Collection Principles
The clip for a rider to pass through from the non-sterile side of the terminus to the unfertile side is merely one portion of the arrival experience for a rider going from an airdrome. Figure 2 places the security theodolite clip in the context of the full reaching procedure for a rider beginning as the rider approaches the airdrome belongings from the main road.
The public presentation of many different average transit systems are routinely evaluated utilizing quantitative informations ( 1 ) . For illustration, for over 50 old ages, travel clip along main roads sections has been evaluated by human perceivers entering the clip and licence home base Numberss of vehicles go throughing two or more locations and so later ciphering the travel clip between the two Stationss by taking the difference in the observation times. More late, transit bureaus have begun leveraging the extended incursion of nomadic electronic devices with alone ascertainable Bluetooth identifiers ( 2, 3, 4 ) . These techniques provide a really economical mechanism to roll up the big informations sets necessary to do system direction determinations based upon sound statistical informations ( Mistake: Reference beginning non found ) . These Bluetooth informations aggregation techniques have besides been adopted for usage in airdrome environments ( Mistake: Reference beginning non found ) . In both the main road and airdrome applications about 5 % to 10 % of the vehicles or riders have ascertainable identifiers that can be used for roll uping informations ( Mistake: Reference beginning non found, 7 ) . Although this is a comparatively modest sample size, it is sufficiently big to supply statistically robust word picture of system public presentation ( Mistake: Reference beginning non found )
Figure 3 illustrates the rules of utilizing two Bluetooth monitoring Stationss at George Bush Intercontinental Airport ( IAH ) to roll up alone identifiers associated with Bluetooth devices as they pass the mark bridges on John F. Kennedy Blvd ( Station A ) and William Clayton Pkwy ( Station B ) . To cipher the travel clip between two different Bluetooth Stationss, the difference in clip recorded from alone devices is calculated. For illustration, say a vehicle passed Station A at 9:55am and Station B at 9:59am. The elapsed travel clip between the two Stationss was calculated to be 4 proceedingss. For lucidity, this illustration rounded clip to the nearest minute, but the clip casts of the alone identifiers are in fact logged to the nearest 1 2nd. This individual informations point by itself has small value. However, by consistently analysing this information over a twenty-four hours ( Figure 3b and c ) or multiple yearss ( Figure 3d ) tendencies and forms can be identified. For illustration Figure 3d, shows a spread secret plan of about 5800 Bluetooth identifiers matched over the two hebdomad period between 28 August and 12 September. Figure 3c shows a spread secret plan for merely September 4, 2010. It is clear from both Figure 3c and Figure 3d that most “ through ” vehicles take 4-5 proceedingss to go from Station A to Station B. Both Figure 3c and Figure 3d have a maximal value of 30 proceedingss on their Y-axis.
Figure 4b provides a statistical word picture of all vehicles that spent 90 proceedingss or less going from Station A to Station B ( 267 vehicles ) on September 4, 2010. In this instance about 40 % of the vehicles had transit times between Station A and B of under 5 proceedingss ( Figure 3 ( I ) ) . Those vehicles with theodolite times under 5 proceedingss were likely through vehicles that did non halt at the terminus.
Sensor Locations for Qualifying the Passenger Arrival Experience
Figure 4 shows the location where nine ( 9 ) Bluetooth supervising Stationss were installed at IAH during August and September 2010.
Stations A and B were located on John F. Kennedy Blvd and William Clayton Parkway severally to gaining control identifiers associated with riders come ining the airdrome belongings.
Stations E and G were located next to the vehicle entrances at the Terminal C parking construction.
Station C was located in the non-sterile side of Terminal C
Stations D, F, H, and M were located on the unfertile side of Terminal C.
Data Reduction Principles
During the two hebdomad informations aggregation period ( 29 August to 12 September ) about 5.5 million clip stamped identifiers were collected at 9 different Stations ( Figure 4 ) . This resulted in over 200,000 travel records between the assorted Stationss ( Table 1 ) . With this volume of informations, it is of import to develop systematic statistical processing techniques and visual image tools to measure system public presentation and place chances for betterment. By plotting a clip history of checkpoint theodolite times from the non-sterile side to the unfertile side of the Terminal C checkpoint a huge sum of information is shown ( Figure 5 ) .
Looking at Figure 5, it is really difficult to find which yearss had the longest delay times by review. However, by ciphering the 25th, 50th, and 75th percentile theodolite times and plotting them in Figure 6, that graphic provides a better mechanism for comparing day-to-day security theodolite times. From Figure 6 ( two ) , we can see the average delay clip on the Tuesday after Labor Day ( 9/7 ) was about 17.5 proceedingss. We can besides see that the fastest 25th percentile transited the security line in about 10 proceedingss. The 25th percentile is likely a good index of security theodolite times for frequent flyers. In contrast, we see the 75th percentile theodolite times ( Figure 6 ( two ) ) are about 26 proceedingss and likely is representative of less experient travelers. Comparing Figure 6i with Figure 6ii, it appears that the security theodolite times associated with the Tuesday rush after Labor Day ( Figure 6ii ) were longer than the security theodolite times during the period before the Labor Day weekend. We believe the quartile representation shown in Figure 6 makes it easier to see these tendencies so the spread secret plan representation shown in Figure 5.
Data Fusion Concepts
Although Figure 6 is utile to place gross day-to-day theodolite clip tendencies, it gives small penetration into how staffing degrees might be adjusted within a twenty-four hours. To derive insight into checkpoint staffing, it is desirable to measure:
Volume of riders being screened by hr
Quartile secret plans qualifying security theodolite times by hr
Number of security lanes open in a given hr.
Figure 7 shows a day-to-day checkpoint operations study ( COR ) developed in this survey that diagrammatically characterizes all of this information. In this peculiar illustration, Figure 7a shows the checkpoint volume for both the North and south lanes by hr. Figure 7b shows the quartile secret plans for rider security theodolite times by hr. Figure 7c shows the figure of lanes unfastened by hr. Upon review of this secret plan, it is clear that the theodolite clip through the checkpoint increases significantly during the 10:00 period ( Figure 7b, Callout I ) , with the average checkpoint theodolite clip increasing to about 25 proceedingss. Figure 7c indicates that one of the south lanes was shut down during the 10:00 period ( Figure 7c callout two ) , yet the rider volumes during the 10:00 period remained about every bit strong as the 9:00 period ( Figure 7a, callout three ) . Obviously there are issues such as testing staff remainder interruptions that need to be considered, but this checkpoint operations assessment study provides a good illustration of how passenger theodolite clip informations can be fused with checkpoint volume and unfastened lane informations to quickly place chances for operational betterments. Although this instance demonstrates an chance for betterment, it is an stray catch shooting of Terminal C and does non see the holistic operation of all nine checkpoints that must be staffed at IAH. Consequently, this illustration should non be considered a review of the staffing degrees by the TSA.
Figure 8 illustrates an illustration where possibly there is an chance to apportion the same staffing degrees, but at different hours. In this peculiar illustration, Figure 8b ( callout four ) shows a average delay clip during the 16:00 period nearing 20 proceedingss and merely two lanes unfastened ( Figure 8c, callout V ) . In contrast, during the 18:00 period, when the average delay clip is below 10 proceedingss ( Figure 8b, callout I ) , three lanes are unfastened ( Figure 8c, callout two ) and the rider checkpoint volume is rather low ( Figure 8a, callout three ) . This is an illustration where alternatively of staffing the checkpoint with three lanes during the 17:00 and 18:00 periods, the checkpoint would likely be more expeditiously operated by staffing three lanes during the 16:00 and 17:00 periods.
In Figure 9, the delay times during the 7:00, 8:00, 9:00, and 10:00 periods ( Figure 9b, callout I ) all exceed 20 proceedingss. Figure 9c shows that merely three lanes are unfastened during this period, with comparatively high checkpoint volumes ( Figure 9a, callout three ) . Although three lanes is likely an appropriate staffing degree for a regularly Saturday, this peculiar twenty-four hours is the Saturday of Labor Day weekend and the checkpoint likely would hold benefited from holding all four lanes open during this clip period.
The information represented in these checkpoint operations studies ( COR ) can besides be used to calculate rider delay clip costs. Table 2 shows the estimated person-hours of hold for Friday September 3, and Saturday September 4, 2010. Using an recognized mean pecuniary value for a rider ‘s clip, the economic impact of delay times can be assessed and used by directors to do investing determinations sing appropriate checkpoint staffing degrees and operations.
Asessing the Entire Passenger Arrival Experience
Figure 2 illustrates that there are many activities associated with a rider geting at an airdrome. Although riders, often make remarks about how long they spend waiting in line to pass through security, it is enlightening to set that in the context of the full reaching experience. Figure 10 shows the Accumulative Frequency Diagrams ( CFDs ) for the elapsed clip between an geting riders go throughing the mark span on John F. Kennedy Blvd ( Figure 4 Station A ) and
Arriving at Terminal C ( non-sterile side )
Uncluttering Terminal C ( unfertile side, recompose countries )
Arriving at the North or south wings of Terminal C ( Figure 4 Stations M or F )
These CFDs are for a four hr period, between 0600 and 1000 on Friday, September 3, 2010. For the clip period characterized in Figure 10, we can see that the fastest anybody makes the trip from Station A to the non-sterile part of Terminal C is about 5 proceedingss ( Figure 10, callout I ) and likely corresponds to person being dropped off. The average clip for the trip from Station A to the non-sterile part of Terminal C is about 9 proceedingss ( Figure 10, callout two ) and about 11 minute longer to pass through security and enter the recompose country ( Figure 10, callout three ) . We can besides see the average clip from the recompose country to the North or south wings of Terminal C ( Figure 4 Stations M or F ) is about 2 proceedingss ( Figure 10, callout four )
This paper reports on a survey utilizing Bluetooth detectors to quantitatively measure the operations of airdrome security checkpoints. Bluetooth detectors were deployed at George Bush Intercontinental Airport from 28 August to 13 September, 2010 and collected over 200,000 travel clip records ( Table 1 ) between nine Stationss located along entree roads, parking garages, non-sterile country of terminuss, and unfertile country of terminuss ( Figure 4 ) . In add-on, hourly checkpoint rider volumes were collected. The overpowering volume of this information necessitated the development of statistical word picture processs and artworks that characterize both the travel clip records and checkpoint information. Several interesting observations and recommendations were derived from this survey:
Approximately 40 % of the vehicles going along the path from John F. Kennedy Blvd to Will Clayton Parkway transited in less the 5 proceedingss and probably did non halt at the terminus ( through traffic ) .
Mentioning to Figure 10, the average travel clip between when a rider enters airport belongings and arrives at Terminal C is about 9 proceedingss ( 75th percentile is about 13 proceedingss )
The merger of the checkpoint volume, figure of lanes unfastened, and wait clip in a graphical format is an improbably powerful direction tool. These checkpoint operations studies ( COR ) provide dashboard degree studies utile for direction to quickly measure operations and place chances for improved allotment of staffing. Figure 7 – Figure 9 were used to propose several chances where at that place appeared to be chances to cut down security checkpoint delay times.
In add-on to measuring the average delay times, the quartile representation of theodolite clip shown in Figure 7 – Figure 9 provide a mechanism for gauging the delay times for frequent travellers ( 25th percentile ) and less experient travelers ( 75th percentile ) .
The checkpoint volume and delay clip statistics provide the necessary foundation for calculating quantitative user costs associated with delay times on different yearss. Table 2 shows an illustration of how these user costs can be compared between two different yearss and organize the footing for doing economic tradeoffs between staffing degrees and user costs.
Longer term, it is recommended these informations aggregation and informations merger techniques be considered for lasting deployment airdrome broad so that operations of checkpoints at different terminuss can be compared and possibly distributed in real-time to directors and the populace so they can utilize that information to do real-time determination.