Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 196 (2017) 309 – 314

Creative Construction Conference 2017, CCC 2017, 19-22 June 2017, Primosten, Croatia

What Authentication Technology Should Be Chosen for

Construction Manpower Management?
Sangyoon Chin, Ph.Da*, Inchie Kimb, Cheol-Ho Choi, Ph.Dc
a
Professor and Chair, School of Civil, Architectural Engineering & Landscape Architecture, Sunngkyunkwan University, Suwon, 16419, S. Korea
b
Research Assistant, Dept. of Convergence Engineering for Future City, Graduate School, Sungkyunkwan University, Suwon 16419, S. Korea
c
CEO, DoallTech Co. Ltd, SK V1 Center, 11, Dangsan-ro 41-gil, Yeongdeungpo-gu, Seoul, 07217, S. Korea

Abstract

The objective of this paper is to analyze the feasibility and limits of various authentication technologies such as radio frequency
identification (RFID), quick response (QR) codes, and fingerprint, vein, iris, and facial recognition for manpower management and
access control at construction sites. Based on the analysis of limits and features of authentication technologies, requirement analysis
in construction field operations through interviews and site visits, and video study based on construction workers’ actual use of the
technologies, this study identified that the false rejection rate (FRR) and the processing time (PT) per person of an authentication
technology are among the governing factors for making decisions to adopt an authentication technology for manpower management
and access control. This study developed an equation to estimate the total process time using FRR and PT for each of the
technologies, and the estimation results show that the difference in total process time can be up to 4 − 5 times larger for each
technology. This can have a significant effect on the number of units to be installed at the level of about 2,000 or more construction
workers.

©©2017
2017TheTheAuthors. Published
Authors. by Elsevier
Published Ltd. This
by Elsevier Ltd. is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the Creative Construction Conference 2017.
Peer-review under responsibility of the scientific committee of the Creative Construction Conference 2017

Keywords: access control; authentication technology; manpower management; RFID; QR Code; Biometrics

1. Introduction

Manpower management is one of the most critical jobs and should be handled on a daily basis at construction sites
since manpower information can be utilized for access control, productivity analysis, labor payroll, progress, and

* Corresponding author. Tel.: +82-31-290-7568; fax: +82-31-290-7570.

E-mail address: schin@skku.edu

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the Creative Construction Conference 2017
doi:10.1016/j.proeng.2017.07.204
310 Sangyoon Chin et al. / Procedia Engineering 196 (2017) 309 – 314

safety management. A labor-intensive construction project attracts tens of thousands people every day, and manual-
based manpower management is practically useless in such a condition. Meanwhile, some construction projects
require high-level security control at construction sites.
Currently, various authentication technologies, such as radio frequency identification (RFID), quick response (QR)
codes, and fingerprint, vein, iris, and face recognition, are adopted for manpower management and access control at
construction sites. However, due to a lack of understanding of authentication technologies and their applications in
construction environments, limits in adopting such technology for manpower management have been discovered. In
addition, there has been a lack of research on the feasibility and adoptability of various authentication technologies at
construction sites from a construction manager’s point of view. Further verification and validation of these
technologies based on the requirement analysis of construction sites are necessary for successful and practical
application in manpower management and access control.
Therefore, the objective of this research is to develop a method to evaluate the feasibility and limits of various
authentication technologies for manpower management at construction sites. Considering the features and
requirements at construction fields, this study analyzes characteristics of the authentication technologies of RFID, QR
codes, and various biometric recognition technologies such as fingerprint, vein, iris, and face recognition.
Requirements of construction field operation, limits and problems, and decision-making factors for authentication
technologies are derived based on site visits, interviews with construction practitioners, and video study. Finally, a
guide framework consisting of consideration factors and an equation to estimate the total process time and number of
units required for access control are proposed.

2. Authentication Technologies

2.1. Radio Frequency Identification (RFID) & Quick Response (QR) Codes

RFID is a touch-less, non-line of sight, electromagnetic spectrum-based technology consisting of a reader and tags
[1]. Passive tags are typically used for construction manpower because of their low cost for tags compared to active
RFID tags. Passive RFID cards are assigned to each construction worker so that they scan their cards at the gate of a
construction site for access control and time tracking of working hours. Due to its touch-less feature for recognition,
RFID supports fast identification of workers and is particularly effective for a construction site that involves a large
number of workers. For example, a Samsung E&C-adopted RFID-based manpower management system can handle
12,000 workers at its peak time [2]. However, using RFID can provide a burden to managers at the construction site
since they have to issue cards to construction workers and manage lost cards.
The features of QR code, a type of 2D barcode, are described as follows [3]. QR codes are easy to generate and
can store much more information (maximum 7,089 characters) than barcodes. As QR codes are resistant to dirt and
damage, data in QR codes can be restored even if the symbol is partly damaged or dirty. Further, they support high-
speed reading from 360-degree directions and can be printed on paper or viewed as images on a smartphone. Due to
these features, it is easy to generate QR codes with embedded essential information on a worker such as name, birth
year, and blood type. QR codes are also inexpensive since they can be printed on paper or sent to a worker’s
smartphone. When there is a new worker or a visitor at a construction site, a new QR code can be immediately
generated and printed for that person. However, using QR codes under direct sunshine should be avoided because they
are a line of sight reading technology using a camera or scanner. In addition, QR codes can be easily copied and so
are subject to false reporting.

2.2. Biometric Recognition

Various types of biometrics have been developed and applied, even in the construction industry, and they include
fingerprint, vein, iris, and face recognition. These technologies use pattern-recognition and artificial intelligence based
on images of fingerprints, veins, irises, and faces. Therefore, obtaining a good quality image and an algorithm to detect
and recognize a person’s identity are key to the technology [4,5,6].
 Sangyoon Chin et al. / Procedia Engineering 196 (2017) 309 – 314 311

Fingerprint recognition is a well-known and widely used technology. Vein recognition uses three types of veins,
finger vein, palm vein, and dorsal hand vein, and it uses a near-infrared LED light and charge coupled device camera
[7]. Iris recognition uses video images of one or both irises of the eyes, which are unique and well-protected against
damage and wear [8]. Finally, face recognition can use a normal camera to detect a face and recognize various patterns
to identify a person [9].
Because the current technologies of biometrics cannot guarantee perfect recognition, the false acceptance rate
(FAR) and false rejection rate (FRR) are two important indicators in evaluating biometric recognition. The FAR
indicates the possibility of improper acceptance of unauthorized users, while FRR indicates the possibility of improper
rejection of an authorized user [10]. Official tests and studies have shown that fluctuations in FRR values are observed
depending on the developer of algorithms for biometric technologies. Waston et al. [5] showed that the best FRR in
fingerprint recognitions was 0.0197, while the worst was 0.2995; Bharathi et al. [11] showed that the best FRR in vein
recognition was 0.05, while the worst was 4.8; Phillips et al. [6] showed that the best FRR in iris recognition was
0.014, while the worst was 0.038; and Grother and Ngan [4] showed that the best FRR in face recognition was 0.041,
while the worst was 0.205.

3. Requirement Analysis for Manpower Management and Access Control

Through site visits and interviews with construction practitioners where the authentication technologies mentioned
above have been adopted, the issues to be considered were derived as follows. The most critical factors in manpower
management and access control are accuracy and time. Authentication technology should be able to accurately and
quickly handle hundreds to more than 10,000 construction workers on a daily basis, because the recognition of a
construction worker is directly related to payroll, productivity, progress, safety, and many others factors. In addition,
the waiting time should be minimized during access control. Based on the experience of the authors, the maximum
allowable waiting time for site entrance through access control should be 30 minutes. Some practitioners insist that
gate passage for all workers be handled within 15 minutes so that they can be in working mode within 30 minutes.
Through on-site monitoring, it was observed that reading errors and processing time for recognition of RFID and
QR codes were superior to those of biometric recognition. However, practitioners pointed out that RFID or QR codes
could be used for improper applications. An example is that one person can recognize several peoples’ ID cards and
be deceived, under the impression that more people are input than the actual number involved. With regard to
processing time, the average time using RFID cards was 1.2 sec/man, while the time for QR codes was 1.4 sec/man.
The authors observed that adjusting the QR code position relative to the light of the QR scanner causes differences in
the process time.
Construction practitioners pointed out that biometrics have the disadvantages of comparatively higher error rate
and longer time for detection and recognition. However, they also have various advantages, including that construction
laborers do not have to carry ID cards once they are registered and higher levels of security and accuracy for manpower
allocated at a site. That said, biometrics-based errors at the construction site were mostly FRR-related errors, and a
much higher error rate was noted compared to previous studies. As Choi [12] pointed out, the scratched or worn
fingerprints of construction workers have caused recognition problems when using fingerprint recognition technology.
Furthermore, high error rates were observed in a certain construction site where a vein recognition technology had
been adopted so that the site finally abandoned this technology. Users at the site presume that the influence of
construction works or weather could have caused changes in vein condition so that the technology interpreted the
same person’s vein as belonging to a different person. Contaminants such as cement and paint on fingers and hands
were found to also have a significant effect on recognition error rate. In addition, some sites adopted a mixed system
that uses biometrics and a password pad or RFID/QR code in order to minimize the search time to find the detected
biometrics from the manpower database.
Overall, biometric recognitions have been observed to take more time because of the need for a specific pose for
accurate recognition; the process time took two to four times longer than those using RFID or QR codes, as shown in
Table 1. In the cases of fingerprint and vein, additional time was needed to achieve the correct position to avoid false
rejection. With iris and face recognition, the worker had to adjust the camera height or bend his/her worker’s so that
the camera could detect the iris/face, increasing the procedure time. Furthermore, facial expression had a great impact
312 Sangyoon Chin et al. / Procedia Engineering 196 (2017) 309 – 314

on recognition error. The performance test performed by the US NIST [4, 13] shows that the FRR when using a web
camera for face recognition is greater than 11% and increases further when considering wild shot photos with various
facial expressions. During the winter, many construction laborers prefer to wear hats and masks, resulting in decreased
performance of face recognition technology. Finally, iris and facial recognition equipment have a significantly lower
recognition rate in the presence of backlighting, which means that it is necessary to implement a block for sunlight,
resulting in additional costs.

4. Total Process Time and Units for Manpower Management and Access Control

From the requirement analysis above, consideration factors were drawn up for adopting recognition technology
for manpower management and access control. There are pros and cons for each recognition technology. One
construction site might consider security the most important factor, while another site might prioritize processing time.
Unfortunately, it seems that there is no single technology that can satisfy all such needs. However, the check-in process
for all construction workers should be completed within a limited time at a reasonably accurate level.
Therefore, an equation to estimate the total processing time required for access control on the basis of maximum
daily number of construction workers was derived in this study. The maximum number of daily workers at DoallTech,
where one of the authors is in charge, was 16,000 laborers, and the allowable maximum waiting time was 30 minutes.
Considering the processing time and error rate of each recognition technology, the total processing time to process the
maximum number of construction laborers for the target site can be estimated. Dividing the total time by the allowable
maximum waiting time yields the required number of units needed to process the maximum number of workers within
the maximum allowable time period.
To derive the average processing time per person, a video study on the access control of the construction sites using
each recognition technology was performed, and the results are listed in Table 1. In addition, the authors showed that
the time required for re-recognition due to errors would increase by three fold per person based on the video study.
Since biometrics have not been widely introduced in the construction industry, the published FRR values described in
Section 3.3 were used as the error rate, assuming that the best available recognition technologies for each biometrics
approach are adopted. The reason for using only the FRR value is that most errors in the field are not FAR, but are
instead related to FRR, which results in false rejection. Also, it was found that the time needed to pose for correct
detection of biometrics was larger than the time wasted due to recognition errors. Since there was no construction
field adopting vein recognition at the time of the study, the processing time for veins was assumed to be the same as
for fingerprints. Finally, this study developed an equation to estimate the total processing time due to recognition
technology, given by Eq. (1), and the results are shown in Table 1. Table 2 shows the number of units required to
process the maximum numbers of workers within 30 minutes.

ܶ௜ ሺ‫ܿ݁ݏ‬ሻ ൌ  ሺͳ ൅  ‫ܴܴܨ‬௜ ή ͵ሻ ή ‫ܯ‬௠௔௫ ή ܶܲ௜ (1)

5. Conclusion

Accurate and prompt identification is essential for manpower and access control at a construction site, particularly
for large-scale building projects. In this paper, requirements and limitations considering the characteristics of each
technology and the conditions of construction sites to adopt authentication technology are derived based on a literature
review, site visits, interviews, and video studies. Based on the analysis, this study developed an equation for estimating
the total processing time and the amount of equipment per recognition technology that has the greatest effect on
decision-making for selecting an appropriate technology for a given site.
Through the video study and total process time estimation shown in Table 1, it was found that the difference in
recognition time can vary by up to 4 − 5 times for each technology. This can have a significant effect on the number
of units to be installed, as shown in Table 2, at the level of about 2,000 or more construction workers. In addition, it
was found through the video study that the time spent to pose for recognition by the biometric device is much larger
than the time wasted due to rejection errors and has the greatest effect on the total time. Therefore, an automatic
 Sangyoon Chin et al. / Procedia Engineering 196 (2017) 309 – 314 313

adjustment function for devices needs to be developed to reduce the time required for posing for recognition,
considering the conditions of construction sites.
The equation presented in this paper needs further development to accurately predict the total process time per
recognition technology because there have not been a sufficient number of field sites using biometrics technology,
and even the same type of biometrics can have large FRR variances according to the algorithm used. Nevertheless,
the consideration factors and equations presented in this paper could be used as a framework to compare the
characteristics of each technology and consider the expected installation environment and cost.

Table 1. Total Processing Time of Authentication Technology

Processing Maximum Number of Workers at Peak Time (Mmax)

Time Per
Technology FRR 100 500 1,000 2,000 5,000 10,000
Person
(TPi, sec)
Total Processing Time (Ti, sec)
RFID 1.20 0.005 122 609 1,218 2,436 6,090 12,180

QR codes 1.40 0.005 142 711 1,421 2,842 7,105 14,210

Fingerprint 2.67 0.019 282 1,411 2,822 5,644 14,111 28,222

Vein 2.67 0.050 307 1,535 3,071 6,141 15,353 30,705

Iris 3.53 0.014 368 1,839 3,678 7,357 18,391 36,783

Face 5.05 0.041 567 2,836 5,671 11,342 28,356 56,712

Table 2. No. of Units Required to Process Workers at Peak Time

Mmax
Technology 100 500 1,000 2,000 5,000 10,000
No. of units (based on allowable max. waiting time 30 min)
RFID 0.07 0.34 0.68 1.35 3.38 6.77
QR codes 0.08 0.39 0.79 1.58 3.95 7.89
Fingerprint 0.16 0.78 1.57 3.14 7.84 15.68
Vein 0.17 0.85 1.71 3.41 8.53 17.06
Iris 0.20 1.02 2.04 4.09 10.22 20.43
Face 0.32 1.58 3.15 6.30 15.75 31.51

Acknowledgements

This research was partially supported by the Ministry of Science, ICT, and Future Planning in Korea through
research grant R7122-16-005 titled “Development of global market-oriented resource management for construction
site management.” This research was also partially supported by the Ministry of Land, Infrastructure, and Transport
in Korea through the “U-City Master and Doctor Course Grant Program.”

References

[1] Shepard, S. (2005). RFID: radio frequency identification. McGraw Hill Professional.
[2] Kim, S. A., Chin, S. Y., Jang, M. S., Jung, C. W., & Choi, C. H. (2015). A Development of Framework for Selecting Labor Attendance
Management System Considering Condition of Construction Site. Korean Journal of Construction Engineering and Management, 16(4), 60-69.
314 Sangyoon Chin et al. / Procedia Engineering 196 (2017) 309 – 314

[3] Denso Wave Incorporated (2017). What is a QR Code? <http://www.qrcode.com/en/about/> (Jan. 30, 2017)
[4] Grother, P., & Ngan, M. (2014). Face Recognition Vendor Test (FRVT). Performance of Face Identification Algorithms.
[5] Watson, C., Fiumara, G., Tabassi, E., Cheng, S. L., Flanagan, P., & Salamon, W. (2014). Fingerprint vendor technology evaluation. NIST
Interagency Report 8034.
[6] Phillips, P. J., Bowyer, K. W., Flynn, P. J., Liu, X., & Scruggs, W. T. (2008, September). The iris challenge evaluation 2005. In Biometrics:
Theory, Applications and Systems, 2008. BTAS 2008. 2nd IEEE International Conference on (pp. 1-8). IEEE.
[7] Yuhang Ding, Dayan Zhuang and Kejun Wang (2005). "A Study of Hand Vein Recognition Method." Institute of Electrical and Electronics
Engineers, vol.4, pp. 2106-2110.
[8] Hosseini, M. S., Araabi, B. N., & Soltanian-Zadeh, H. (2010). Pigment melanin: Pattern for iris recognition. IEEE Transactions on
Instrumentation and Measurement, 59(4), 792-804.
[9] Sato, A., Imaoka, H., Suzuki, T., & Hosoi, T. (2005). Advances in face detection and recognition technologies. NEC Journal of Advanced
Technology, 2(1), 28-34.
[10] Thakkar, D. (2016). False Acceptance Rate(FAR) and False Recognition Rate (FRR) in Biometrics, Bayometric,
<https://www.bayometric.com/false-acceptance-rate-far-false-recognition-rate-frr/> (July 7, 2016).
[11] Bharathi, S., Sudhakar, R., & Balas, V. E. (2015). Hand Vein-based Multimodal Biometric Recognition. Acta Polytechnica Hungarica, 12(6),
213-229.
[12] Choi, C. H. (2004). “A case study and future directions of Radio Frequency Identification system”, Proceeding of Annual Conference of the
Construction Engineering and Management Institute of Korea, 5, pp. 145-152.
[13] Ngan, M., & Grother, P. (2015). Face recognition vendor test (FRVT) performance of automated gender classification algorithms. National
Institute of Standards and Technology.