An RTLS-Based Approach to Cyber-Physical Systems Integration in Design and Construction

There have been several approaches to integrating physical construction components and their virtual models using RFID tags. These enable the movement of components to be tracked on the construction site. However, there is inadequate support for bidirectional coordination between these components and their virtual representations. Also, these approaches often involve manual input of status information into the tags and do not support tracking the permanent installed position of tagged components for consistency maintenance between the as-built and the as-planned models. As such, there are difficulties with ensuring accurate and timely updating of building information models and tag information during the construction process. A major bottleneck in achieving this integration is the choice of appropriate mechanisms for binding physical components with their virtual representations. This paper presents an approach to facilitating bidirectional coordination between physical construction components and their virtual models. Specialized real-time location sensing (RTLS) tags were used for tracking the position and status of physical construction components. This approach showed significant opportunities for enhancing real-time construction consistency checking, which will aid proactive decision making and control. The paper also discusses experiments undertaken to demonstrate the proposed RTLS-based system and highlights the merits and demerits of adopting the proposed approach on a full-scale project.


Introduction
Being able to accurately and efficiently monitor construction progress in real time enables project managers to detect schedule delays early and make corrective decisions [1].Over the years, virtual models have proved their worth in construction progress monitoring as a means of visualizing the construction process.Virtual models also contain virtual representations of building components which can be linked to their physical representations on the construction site.As a facility moves through the life cycle from planning to design to construction and to facility management, some information can be embedded in the virtual building components and this provides another integrated database of relevant information that can be used by the project team during the construction and postconstruction phases of the constructed facility.This potential of virtual models is evident in the research by Anumba et al. [2], Sørensen et al. [3], and Motamedi and Hammad [4] which identified that integrating virtual models and the physical construction can improve the information and knowledge handling from design to construction and maintenance, hence enhancing control of the construction process/constructed facility.
A number of researchers [4,5] have investigated the integration of virtual models and the physical construction for progress monitoring using radiofrequency identification (RFID) tags.However, these approaches still involve manual input of status information (such as installed or uninstalled) into the tags and do not enable tracking the permanent installed position of tagged components on the construction site.As such, there are difficulties with ensuring accurate and timely updating of building information models during the construction process.A more effective approach to construction progress tracking will require automatically tracking the positions and status of an installed/uninstalled component without manual input to the tags.The need for such automated approach to construction has long been identified by Navon and Sacks [6] who cited that construction personnel spend a significant amount of the time on recording and analyzing site data and Cheok et al. [7] who stated that 2% of construction work is devoted to manual tracking and recording of progress data.Data collected using manual methods are usually not reliable or complete due to the reluctance of the workers to record these data, hence the need to reduce data collection time, while providing project managers with timely and accurate information.
Being able to track the permanent installed positions of tagged components enhances consistency maintenance between building information models and the physical construction.Also, there is a scope for the use of these models for enhancing the real-time bidirectional coordination between the design team and personnel on the construction site [8].This will enable documentation of as-built information which is necessary for lifecycle management of the constructed facility as well as serving as a basis for the active control of installed components such as light fixtures [9].An effective way to enable consistency maintenance while enhancing bidirectional coordination between BIM and the physical components is using a cyber-physical systems approach.In the context of this research, the term cyber-physical systems approach is taken to mean a tight integration and coordination between virtual models and the physical components.
A cyber-physical systems approach to integrating the virtual models and the physical construction (for consistency maintenance and bidirectional coordination-Figure 1) requires the use of specialized real-time location sensing (RTLS) tags capable of both position sensing and data storage capability.These RFID tags are real-time location sensing tags capable of tracking the position of tagged components and storing information for integrating virtual models and the physical components.This paper describes an approach to cyber-physical systems integration of design models and the physical construction using RTLS tags.It starts by describing the challenges with current integration efforts.The paper also describes the cyber-physical systems approach and system architecture which brings together, the key enabling technologies adopted in the proposed approach.

Challenges with Current Component Level Integration Efforts
A number of researchers have attempted to integrate virtual models and physical construction using RFID tags (passive and active tags).These technologies are used for identifying, tracking, and locating materials, vehicles, and equipment in the construction industry.The following research efforts have focused on the use of these technologies for progress and lifecycle management in construction projects: Chin et al. [5] examined the utilization of 4D CAD and passive RFID tags for progress management of steel works (from the manufacturers yard to installation).The authors developed a system which renders virtual components in the 3D CAD model according to the as-built status (such as manufactured, inspected, installed, and uninstalled) of tagged steel components.In their approach, to track the status of tagged steel components, the passive RFID tags are scanned and the status information is manually embedded in the tags using a personal data assistant (PDA).The use of passive RFID tags requires scanning the tags at close range and the entry of installation status highly depends on worker's motivation.If a component is mistakenly left out or unscanned before being installed, the status will not be updated in the model and, thus, affects the accuracy of the model.Also, this approach does not address how changes can be managed during the steel erection process.For example, if after the steel fabrication, a mistake is identified that requires a change in the erection process, there is no way to communicate the affected components to the steel erector on the job site.Also, the approach does not address some temporary constructability issues which usually occur on the job site such as if a large mechanical equipment needs to be lowered into the building later during construction and this requires some steel members to be left out, there is no way to communicate the affected members to the structural engineer (who evaluates the remaining structure for stability before the equipment is brought into the building).Hu [10] also investigated the integration of passive RFID and fourdimensional computer-aided design (4D CAD) for tracking the status of construction components (such as pipes, equipment, steel columns, and beams).Construction components are tagged with RFID passive tags and an RFID reader is used to track their status from the manufacturing or fabrication plant to the construction site where they are installed.Once the components are installed, construction personnel use a personal digital assistant (PDA) with an embedded RFID reader to manually embed status information into the tags.This status information is captured in a 4D CAD model of the initial project schedule for comparisons between actual and initial work schedules.The use of passive tags makes the approach unsuitable for project sites with a large number of components as there are possibilities that certain components may be missed out.Also, since passive tags can only be identified at close range, it might be difficult locating/tracking components on job sites.
While the aforemantioned authors utilized passive RFID tags, Motamedi and Hammad [4] investigated the integration of physical components and BIM for lifecycle management of facility components using active and passive RFID tags.In this approach, the authors proposed permanently attaching RFID tags to facility components where the memory of the tags is populated with BIM information taken from a BIM database.The BIM data stored on the tags provides a distributed database that allows access to different players who do not have real-time access to a central database.The authors demonstrated the feasibility of this approach using two case studies of construction progress and lifecycle management of construction components.However, there is no way to track the actual placement of the installed or uninstalled components using the active RFID tags; that is, if the components are installed in the wrong location, there is no way of tracking this in the model.The authors used active RFID tags with memory storage, which can be used to link the physical components with their virtual representation in the model.But this was not addressed in this approach: as there is no way to communicate design changes or model updates about components to the personnel on the job site and there is no way to track the status of maintenance of components in the model.
Although, the aforementioned research has demonstrated the potential of RFID tags for integrating virtual models and physical components for progress monitoring and facilities management using RFID tags, the tags cannot automatically track location or placement of tagged components.Tracking placement of installed/uninstalled components is important in progress monitoring to ensure that the right equipment or component is installed in the right place with minimal manual effort.RTLS system has the potential for sensing the location of tagged components.This could be linked to the virtual model for status tracking.Also, the RTLS system has storage capability.Linking the tagged components with their virtual representation in the model will enable real-time communication between the design team and the construction site personnel, thereby enhancing bidirectional coordination.Bidirectional coordination will enable information such as design changes/model updates to be captured by the construction personnel on the job site in real time.This also supports the communication of as-built "status" information to the model.To enable bidirectional coordination, an approach is needed that will tightly integrate the virtual resources and the physical construction/components; this is termed the cyber-physical systems approach.

Cyber-Physical Systems Approach
In the construction context, cyber-physical systems approach is taken to mean a tight integration and coordination between virtual models and physical construction/constructed facility such as to enable bidirectional coordination.The key features of a cyber-physical systems approach are the cyber-physical bridge and the physical cyberbridge is shown in Figure 2 [11].The physical to cyber bridge is the sensing  process, which involves the use of sensing systems to identify, distinguish, locate, and bind the physical components to the virtual representation during construction.On the other hand, the cyber to physical bridge represents the actuation which shows how the sensed information is acted upon by the system.In this context, actuation is taken to mean making control decisions from the sensed information (from the physical to cyber bridge) and/or using the sensed information to physically control the tagged components.

Enabling Technologies.
The key enabling technologies for cyber-physical systems integration such as to enable bidirectional coordination between virtual models and the physical components are discussed as follows.

Real-Time Location Sensing (RTLS)
System.The RTLS system obtained from Identec Solutions consists of real-time location sensing (RTLS) tags (i-Q350 RTLS), RTLS reader (i-PORT M 350 RTLS), satellite nodes (i-SAT 300 RTLS), and an "i-Share" position server.The overview of the RTLS system is shown in Figure 6.
i-Q350 RTLS Tags.The RTLS tags (Figure 3) have position sensing and data storage capability (32 kbyte read/write memory).These tags have a very long communication range up to 500 m (1600 ft) and enable the automated identification, tracking, and tracing of assets and people in areas as large as a steel construction workshop without human intervention.
i-SAT 300 RTLS.The i-SAT 300 RTLS (shown in Figure 4) operates as one of several reference points for the localization of the RTLS tags.The i-SAT 300 RTLS provides information about identity and location of RTLS tags.The i-Q350 RTLS in conjunction with the i-SAT 300 RTLS reference generator allows localization down to a few feet.It operates stand-alone only with power supply.The high rate of transmission allows communication even with rapidly moving tags.
i-PORT RTLS Reader.The i-PORT M 350 RTLS reader, shown in Figure 5, acts as a combination of an RTLS reader (with up to 500 m read/write range) and a satellite node  (capable of about 400 localizations per minute).This means it serves as a reference node as well as an interrogator to retrieve the RTLS ranging data from the tags.When building materials and components are tagged with the RTLS tags, the tags control the communications to the i-SAT nodes.The i-SAT nodes provide position information of the tagged component to the RTLS tag, which then communicates this position information to the RTLS reader.
i-Share Positioning Software.i-Share Edgware is a server application with the primary goal of filtering data to and from the i-PORT M350 RTLS reader installation.The supported features contain handling of various RFID situations like position calculation and sensor data.In addition to this, the server controls the system status and exposes tag communication to business applications.In order to reduce the effort for system integration and avoid typical interface problems like serialization issues, all available interfaces are Web services based.
The location information captured by the RTLS reader is collected in the i-Share positioning software.The i-Share positioning software computes the actual position of that tagged a component with respect to the reference i-SAT nodes and the RFID reader.Likewise information written to the RTLS tags can be captured by the RTLS reader and stored in the i-Share software.The positioning software can be integrated with BIM and other project management applications for as-built documentation and for visualizing progress information.

Mobile Devices.
Mobile devices have long demonstrated opportunities for improving the construction delivery process through providing access to information on site and means of collaboration between project participants.Mobile devices such as tablet PC and PDA have display screen for capturing information about design changes, comments, and relevant information that need to be relayed to the construction team in real time.The construction team can also use the mobile devices for embed information in the tags or directly to the i-Share software to be captured in the BIM in the office.The mobile devices serve as a great tool to communicate back and forth the construction site and the office.In the context of this research, a tablet PC was used.This was used to access model updates and changes.The tablet PC also provided means of embedding information in the tags to be updated in the model.

Communication Network.
The communication network, such as Ethernet and Wi-Fi, plays a great role in enhancing effective integration between BIM and the physical construction.The RTLS reader connects to the positioning server using the Ethernet.The positioning server can be accessed by the project team using the Internet.The mobile devices can also be linked using Wi-Fi to enable information sharing and collaboration between the project team.

System Architecture.
The system architecture is illustrated in Figure 7.The system architecture brings together the key enabling technologies (i.e., virtual prototyping, wireless sensors, mobile devices, and communication network) as a framework for bidirectional coordination between virtual models and physical components.The architecture is based on multiple layers, which are explained as follows.

Sensing Layer.
The sensing layer consists of the RTLS system, comprising the tags, reader, and i-SAT nodes.In this layer, the i-SAT nodes sense the proximity of the tags and send the location data to the tags.The RTLS reader reads the location data from each of the tags and this is captured in the positioning software (where the exact location of the tags is computed).Information can also be written to the RTLS tags from the construction site; this information is captured by the RTLS reader and can be viewed in the model in the office.In this layer, the RTLS system serves three purposes: identification of components, localization of tagged components, and capturing information about the tagged components.

Device
Layer.This layer consists of the client devices (such as the PDA, tablet PC, and mobile phones).This layer serves the purpose of enabling access to the sensed information written to the RTLS tags (from the model).It also enables the construction personnel embed as-built information into the RTLS tags to be captured in the model.

Communication Layer.
This layer contains the Internet and wireless communication networks: wireless personal area networks (ZigBee and Bluetooth), wide area networks and local area networks (which use Wi-Fi to enable access to the Internet).This communication networks connect mobile devices to allow for collaboration and information sharing of the captured RTLS tag information between construction personnel on site.The communication networks also allow the data collected through the mobile devices to be transferred through the Internet to the software and database in the contents and application layer.

Contents and Application
Layer.The contents and application layer contains the database server and the control applications such as the positioning server and project management applications.This layer collects location data from the RTLS tags and computes the exact location of the tags relative to all the devices in space.This layer also stores and is constantly updated with information written to the RTLS tags from the client devices and from the actuation layer.The control applications use the sensed data from the database to make control decisions which can be visualized using the virtual prototype in the actuation layer.

Actuation Layer.
This layer contains the virtual prototype which is accessed through the user interface.The virtual prototype enables the user to visualize how the sensed information (from the contents and application layer) affects the system (showing which components are installed and uninstalled).The human interface enables the user to visualize and monitor the movement of the tags in the space.This layer also enables the user visualize the sensed information from the contents and storage layer.The user can also embed control decisions into the virtual prototype through the user interface to be accessed in the device layer.

Overview of Experiment
In order to carry out the experiment, a prototype system was developed based on the architecture in Figure 7.The steps involved in the prototype system development include the following.

Development of Design Model.
A virtual model of a small-scale building was developed using Autodesk Navisworks.The model serves the purpose of enabling visualization of status of tracked building components and information captured from the project site.The model also enables embedding of model updates or critical information that needs to be communicated to the construction site in real time.Navisworks was utilized because it offers an open.Net application programming interface (API), which enables users to write custom plugins to drive Autodesk Navisworks from outside the graphical user interface (GUI) and automate tasks like changing material properties and embedding information in components.

Physical Model.
A laboratory-scale physical prototype of the Navisworks model was constructed.The physical prototype (Figure 8) consists of nine detachable components.The physical prototype was designed as detachable components to enable easy removal when tracking placement during the experiment.The components of the laboratory-scale physical building prototype were tagged with RTLS tags as shown in Figure 8.

Application Development.
In order to integrate the design model and the physical prototype, two applications were developed using Visual Studio.Net.These applications are described in more detail below.
CPSPlugin.This is the main entry point into Autodesk Navisworks.This plugin was used to invoke the features of Navisworks such as color and property values.This plugin creates and invokes an interface on the design model GUI (Figure 9).This plugin collects "initial" and "changing" location data for each RTLS tag from the database (shown in Figure 11) and compares both data.Based on the results of the comparison, the plugin updates the property of the corresponding virtual component.the tag and writes it to an Access database, to be read by the CPSPlugin.The client application also reads information written to the Access database and writes it to the associated tag.The client application was written using "Windows Forms," in the VisualStudio.Net environment.Using the client application, the user initiates a connection to a Web service interface in Figure 10.The user logins in using his user name and password and the version of i-Share software utilized is confirmed.The user can then select the tagged component he wishes to read from or write to.On selecting the tag ID, he can write in the text box and click on the "Write" button.When the user clicks on the "Read" button, he can also read data written to the tag from the model.

Prototype System.
Figure 11 shows an overview of the developed prototype system.In the figure, the components of a laboratory-scale physical building prototype are tagged with RTLS tags.Each i-SAT node determines the proximity of each tagged component and sends the coordinate information to the tags.The individual coordinate data are read from the tags by the RTLS reader and transferred to the i-Share software.The i-Share software computes the relative distance of each tag in space and sends this to a Web interface.The database is constantly updated with data from the Web interface.A Navisworks plugin (called CPSPlugin) collects the position data from the database and updates the status of the affected elements in the model.Whenever there is a design change or model update, the CPSPlugin captures this change and stores it in a database and in the Web interface where it is received by the i-Share software.The RTLS reader collects this information and writes it to the associated RTLS tag where it can be accessed on site.Also, the tag can be read and updated using a client application installed on the mobile device.The client reads and writes to that tag by connecting to the Web interface.Information written to the Web interface is collected and stored in the Access database where the CPSPlugin updates the associated element with the change.

Prototype System Implementation.
One of the objectives of these experiments was to track when tagged components are installed and uninstalled.The principle adopted here  is that the as-built locations (position coordinates) of the tagged components must be known.These as-built locations are input into the developed CPSPlugin as the tagged component's final "installed" location.These position coordinates are captured in Figure 9 as the "Initial Tag Position."As the tagged components are moved from place, the position coordinates are captured in Figure 9 as the "Changing Tag Position."CPSPlugin compares the "Initial Tag Position" and the "Changing Tag Position" for each tagged component.If these coordinates are the same, the tagged component is considered to be "installed" and the model is updated to reflect this.Conversely, if the coordinates are different, the tagged component is considered "uninstalled" and the model is updated to that effect as shown in Figure 12.
Another objective of the experiment is to demonstrate bidirectional coordination through enhancing communication of critical information between the virtual and physical components.This was achieved using the CPSPlugin.CPSPlugin has read and write features.The read feature enables the capture of tag information from the Access database and writing the information to the associated element in the model.Once CPSPlugin reads new tag information, the element is highlighted and the new information is updated in the TagData property value as shown in Figure 13.
To write to a tag on site, the user writes information to the "Write To Tag" textbox, selects the element (he wishes to write to) in the selection tree, and clicks on the "Write" button as shown in Figure 14.During the indoor test, false movements (multipath effects) were noticed from the RTLS tags while the tagged components were stationary.These false movements resulted in a number of false updates recorded in the model.Figure 15(b) shows the map from the indoor test.The map represents the layout of the test site.This is produced by the RTLS system.The interference was established to be due to a large number of wireless networks in the vicinity of the laboratory.This led to an exploration of outdoor tests.
(The notations T1, T2,. ..,T8 on each green and blue icon represent the tags, with the green tag showing the initial location while the blue tag location shows the final location.The line between the RTLS tags represents the path to movement.)4.6.2.Outdoor Test.The developed system was also repeated in an outdoor environment in a park behind the Engineering Unit buildings at the Pennsylvania State University as shown in Figure 16(a).The system was set up by installing the i-SAT nodes and RTLS reader at the boundaries of the test sites.The i-SAT nodes and the RTLS reader were placed at increased distance apart.The i-SAT nodes were attached to metal poles at a height of 6 m.
The outdoor test was carried out under the following conditions: when the surrounding wireless networks was switched off and switched back on.The surrounding wireless networks were switched off to avoid disruption to the RTLS signals.Under this condition, less multipath effects were noticed as illustrated in Figure 16(b).However, when the  wireless networks were turned back on, more multipath effects were noticed as illustrated in Figure 16(c).
(The notations T1, T2,. ..,T8 on each green icon represent the tags and show that there was no multipath effect at the outdoor site).
(Green and blue icons represent the tags, with the green tag showing the initial location while the blue tag location shows the final location.The line between the RTLS tags represents the path to movement.)

Discussion
The experiment presented in this paper has demonstrated some potential for the use of sensing technologies for tracking actual placement of tagged components, specifically for progress monitoring and capturing as-built information during construction.This is a step further from the existing approach of manually embedding status information into the tags.The experiment was implemented indoors and outdoors.The indoor experiment helped to determine  the suitability of the RTLS system in enclosed sections (e.g., partially completed buildings) during construction and in the constructed facility during the operations and maintenance phase.The outdoor test also helped to the suitability of RTLS system in an open environment such as a construction site for material tracking.The RTLS system showed some potential for real-time location tracking of the tagged components in the outdoor environment.The RTLS system proved more effective and with less multipath movement when deployed outdoors with the wireless signals switched off than when the wireless signals were on.The signals from the RTLS system seem to be disrupted indoors, as a result of the interferences from wireless signals (such as the Wi-Fi, Bluetooth, and ZigBee) within and outside the laboratory.The disruption can be observed from the false movement/multipath effects of the RTLS tags viewed from the indoor site blueprint.However, this disruption relates to tracking the placement of the tagged components.The issue of communicating changes to the job site, and obtaining feedback or as-built documentation in the model proved effective.Changes made on the job site can be written to the RTLS tags to be documented in the BIM.Conversely, the project team can embed notification of changes or alerts in virtual components; this can be captured on the project site (either through the RTLS tag or mobile devices).This process of being able to track and communicate permanent placement of tagged components between the virtual model and the physical construction illustrates the concept of bidirectional coordination.The presented experiment has demonstrated the potential of the RTLS system for bidirectional coordination.Bidirectional coordination is beneficial for access to real-time progress information and decision making.The approach also has great potentials for enhancing as-built documentation which is necessary for lifecycle management of the constructed facility.The outdoor test took place outdoor with walls, trees, and less wireless interference (while the wireless network was switched off and on) compared to the indoor environment.
There is need to extend this work further by investigating other sensing systems and algorithms to enable transitioning of tagged components from outdoor (staging area) to indoor (partially completed site) environment.This is being addressed in the next phase of this research.

Conclusion
The authors presented an investigation into the use of a RTLS system for cyber-physical systems integration of virtual models and physical construction such as to enable bidirectional coordination.This investigation was carried out by conducting two tests (outdoor and indoor) to determine the efficacy of bidirectional coordination between the virtual model representations of building components International Journal of Distributed Sensor Networks   and the physical components themselves.The RTLS system proved more effective outdoors than indoors, as the location coordinates were more stable outdoors than in the indoor situation where high multipath movements were recorded, when the tagged components were stationary.However, despite the multipath effects encountered, the use of RTLS tags for tracking model updates/design changes on the site and also updating the model with as-built information was successful in both the indoor and outdoor environments.This bidirectional coordination is considered very important in the deployment of cyber-physical systems in the construction industry.Thus, there is considerable potential for the application of the RTLS system to other aspects of the construction project delivery process, facility management, and other operations.Also, further work on this project involves investigating other sensors and algorithms to aid continuous capturing of location data as components transition from an outdoor (staging area) to indoor (partially completed site) environment.This will need to be tested on the construction site to identify practical implementation constraints in addition to identifying further benefits.

Figure 2 :
Figure 2: Features of CPS for integrating virtual models and physical building components.

Figure 7 :
Figure 7: Architecture for CPS integration of physical building components and the virtual model.

Figure 10 :
Figure 10: Client interface to display read and write command buttons.

4. 6 .
Test Results.The developed prototype system was tested on two different sites (indoors and outdoors).

4. 6 . 1 .
Indoor Test.The developed system was initially tested indoors in the Intelligent Systems Laboratory in the Department of Architectural Engineering at the Pennsylvania State University as shown in Figure15(a).The i-SAT nodes and RTLS reader were attached to steel columns at a height of 1.76 m.The tagged laboratory-scale prototype was placed on a table at a height of about 0.61 m.

Figure 13 :
Figure 13: Element highlighted and TagData property updated with information.

Figure 14 :
Figure 14: Element selected in selection tree and information in textbox.
Indoor site blueprint showing the i-SAT nodes and stationary tags

Figure 15 :
Figure 15: Map from indoor test for stationary tags.
Figure 16: Map from outdoor test for stationary tags.