interfaces and programs were developed in the proposed system, as
described in more details in Section 5.
5. System development
There are several alternative platforms to develop the proposed
cloud-based system and one of the most popular choices is Google's ser-
vices (e.g. Google Fusion Tables); however, it has limitations when it
comes to developing a cost-effective system for the SMB contractor.
First, Google's platform is experimental, which means they cannot guar-
antee continued support for the platform and their API is frequently
changed. This may force us to reprogram our system frequently in the
near future. Second, there are tight usage restrictions, e.g., on the
amount of queries to the database (e.g. the number of queries per ac-
count is limited to 30 per day), which is not sufficient for the proposed
system. There are other problems such as the authentication issue
which is critical as the proposed system is meant for ubiquitous access
by multiple users. In order to reduce this program complexity and
avoid risks in continuing system operation in the future, we adopted
the Microsoft Azure platform (http://azure.microsoft.com/en-us/). This
is preferred because of many benefits including ease of use for both
end users and developers, a more robust DBMS backend, truly ubiqui-
tous access, and scalability. In addition, its support of SQL DB allows us
to attempt integration with the company's existing barcode system.
Also it is highly affordable with no upfront costs, and the SMB contractor
can keep the system at zero or very low incremental cost, considering
the small to medium data requirements for their business. The overall
architecture on the cloud platform to meet the system and functional
requirements is depicted in Fig. 3.
In each site, there are two redundant RFID readers placed at each
checkpoint in order to increase reading range and readability which
can be degraded particularly when the material is metal [31]. For local
connectivity between the RFID readers and desktops within the site,
the system uses USB network interface cards (NIC) that are bridged to
the thin client via serial connection. The thin client can perform various
tasks including login, connecting RFID readers, tag scanning, and query-
ing product information (see Fig. 4). As discussed in Section 4, this archi-
tecture can be significantly simplified by adopting different RFID
readers that support web application API. The thin client is developed
as a desktop application in the proposed system; however, it could be
developed as a web application using the API, which will enable the
proposed system purely cloud-based.
The cloud system includes the data layer that interacts between thin
clients and the cloud DB for data acquisition. The cloud DB also interacts
with and maintains the data in the barcode system and a user manage-
ment DB. The data stored in the cloud DB are presented by the web front
end to the end users through either desktops or mobile devices. For
example, Fig. 5 shows the spool information page in the web front
end. Fig. 6 presents the schema of the cloud DB that depicts the relation-
ship between tables and their attributes defined for this project. The
main table in the DB is the Spool table, which stores various data on
each spool to be produced and installed, including data for design (e.g.
specification and drawing) and manufacturing (e.g. material and condi-
tions). A number of spools are assigned to a job (project) whose start
and end dates and related documentations are stored in the Jobs table.
The data of users, who are involved in a job, are stored in the Users
table. Once a spool is tagged at the fabrication site for tracking, it is asso-
ciated with an item in the Tagged_Spool table along with the welding
information stored in the Weld table. When the spool is scanned by a
scanner, it is updated in the Scanned_Spool table and the location of
the spool can be tracked by the data of the RFID scanner that are stored
in the RFID_Scanner table. Users can query the status of any spools
stored in the cloud DB and track them through the web front end.
6. System evaluation
6.1. Pre-evaluation
The proposed system with cloud-computing and RFID is expected to
overcome the bottleneck in information flow by providing a platform
for communication across participants and realizing automated materi-
al tracking. In order to assess the benefits of the proposed system over
the existing system, a simulation study was conducted using the
Arena simulation package [32]. Three scenarios were developed and
tested in the simulation: 1) the AS-IS case (the current process with
the information bottleneck described in Section 3 that needs to be en-
hanced); 2) a new workflow with only the cloud-based system without
RFID (manual sca