Our findings (restatement of the problem)
Recall that indoor map data commonly comes in 3D or 2D formats, shown in Tables 1 and 2. For data re-use, these common formats should be output by indoor map creation software (Table 3), and input by indoor map applications (Table 4). However, this is not the case. Indoor map data presently are task, and application-centric. Incompatibility is illustrated in Fig. 1. The point is we found that many indoor map file formats cannot be used widely in location-based service applications.
Problems of interoperability for indoor maps
Indoor map data should be able to be shared among applications, given appropriate data security measures. The Results section of this paper shows that many categories of indoor map creation software cannot supply maps for indoor location-based services applications due to file format incompatibilities. The consequences of this are (1) for most indoor map creation software, an indoor map is the final output—with re-use of map in other applications impossible, and (2) the majority of indoor location-based services generate their own indoor map input. In short, an indoor map might need to be converted or entirely re-worked in order to upload it to different indoor map applications, even within the same application category (see Table 4).
Standardization improves interoperability
Data and metadata standards help ensure that (in this case, indoor map) data can be reused. A standard file format will need to be modified and maintained over time. Some companies both develop proprietary standards and also submit or support committees to promote an open standard [1]. Organizations that develop geospatial data standards include the international Open Geospatial Consortium (OGC), the International Standards Organization (ISO), and the International Hydrographic Organization (IHO), and in the United States, the Federal Geographic Data Committee (FGDC), the American Society for Photogrammetry and Remote Sensing (ASPRS). Of these, however, recommendations for indoor geospatial standards have come from the Open Geospatial Consortium (OGC).Footnote 15
Interoperability would improve if there were fewer file formats to integrate
Surveys of indoor location-based service applications reveal the range of data formats used ([5, 20]. The number of file formats in use could be reduced by de jure or de facto methods. A de jure solution would have some independent consortium or standards body developing open standards and advocating for their use. (Multiple standards development bodies can each develop their own standard and advocate for its use.) A de facto solution would result if some possibly proprietary indoor map applications became so widespread that customers and companies would simply adopt those standards.
To review, interoperability can be achieved either through legislating or through the marketplace. In terms of the standard, both open and proprietary have advantages and disadvantages. Open standards are publicly available and can be altered by developers—in fact, so easily altered that the standard becomes too open and is no longer widely interoperable. Proprietary standards are owned by a company and that company can determines who can use it and whether there is a related cost. Open standards increase the size of the marketplace, whereas proprietary standards increase the share of one firm in that marketplace. Note that open standards often are used in open source software. Open source software is more flexible, given the definition of open source, but it likely requires more effort to use, whereas in general the opposite is true for proprietary software.
Technical solution
Interoperability can be affected not only by data but also at the application or web service level. Using such an application, all formats could translate into a single format. Even so, indoor map data conversion can be inefficient and costly, as well as involve data loss. That is why the focus of this paper is on interoperability at the level of data. Another way to improve interoperability would be to use a data management application to coordinate data use across multiple sites. An example of this for documents (not indoor maps) is the proposed Active Data programming model [18].
Yet another way to improve interoperability would be to re-work applications so that they could accept more data formats. For example, indoor map creation software could provide a 2D file output option such as CAD, and possibly also a 3D output option such as IFC’s Building Information Model. 2D CAD and 3D BIM would be suitable choices because these are so widely used by architects and engineers in the United States. Also, BIM is accepted as input in many indoor map applications, which makes it a good choice in a platform to combine with other data types. CAD is more entrenched in the architecture and engineering industries than BIM, however. Alternatively, XML would be a good choice because of relative ease of editing and viewing, with KML for 2D and X3D for 3D, but it is not used as widely, so it would be more costly to change. Recommendation of XML comes from academia rather than industry [16].
The cost of file format conversion hinders a technical solution. Low-cost file transformation would make it easier to collect the indoor maps that emergency response and other applications need to be practically and commercially viable. But file format transformation between indoor map formats can be time consuming (depending on the formats), and therefore unlikely to be at low cost. Automatic conversion from vector to raster map can be trivial. However, automatic conversion from raster to vector maps can be a problem because some of the data from the raster is lost or transferred incorrectly. Broken lines and lost data may be restored manually, but this process is time consuming. And lots of indoor maps are in raster formats such as JPG and PNG, in which the image is represented pixel by pixel. Without converting a raster format to a vector, a raster map that is enlarged will become pixelated and, at some scale, unreadable. Commercially available geospatial file transformation software includes the FME software by Safe.Footnote 16 FME, however, is not easy to use.
The conversion from 2D to 3D presents technical barriers also. At least one open source tool on GitHub can extrude a 2D building diagram to 3D.Footnote 17 But because most applications that use indoor building diagrams need to show walls inside buildings, extrusion is not efficient because the software does not know automatically which lines indicate walls and which indicate staircases, for example, and walls and staircases extrude differently.
Data-centric solution: De jure body advocating for standards
The Open Geospatial Consortium offers resources such as technical documents, training materials, test kits and reference implementations to help technology developers and users take advantage of these open standards. But also, the OGC promotes standards by providing certification and branding to support products that are compliant. It also runs events where all vendors use the same test data (called testbed events) to increase interoperability among vendor products.Footnote 18 Referring back to Tables 1 and 2, OGC supports the 2D KML and the 3D CityGML and IndoorGML formats. Recently, OGC has put out a specification for large-scale spatial data called 3D Tiles, with the .json extension, that defines a spatial data structure and a set of tile formats.Footnote 19 3D Tiles was designed for rendering huge datasets such as for photogrammetry or 3D buildings, and it can be used with a 3D model or a point cloud.
Data-centric solution: De facto market acceptance of standards
One way that standards become accepted is by widespread use. 2D indoor map formats (see Table 2) dominate presently because most applications use 2D indoor map data as input. Of these, the CAD, KML, SHP and GeoJSON vector formats for indoor maps can be scaled up (that is, transformed to larger scale) and retain map clarity. And of the 2D formats, the CAD formats are used widely in the USA for building diagrams by architects, builders and engineers, while vector formats are suitable also for navigation and localization. Apple is attempting to create a standardized schema for GeoJSON with its Indoor Mapping Data Format (IMDF).
IFC’s Building Information Model (BIM) of the 3D formats has a cross-application update feature that makes it suitable for data sharing. In the Netherlands, 3D models are used predominantly for emergency responder training, whereas 2D models are used in support of actual emergency response operations [23]. The same has been noted in parts of the USA, where Responder training programs use 3D models and actual support of incident continues to rely on 2D.
Implications for data storage
The format in which the data is created, stored, and made available for others is important for data re-use. That is why we discuss data storage options in relation to interoperability. Shared public collections of indoor maps presently in the U.S. are limited in scope, as witnessed by the indoor map database, Google Indoors.Footnote 20 There are relatively few maps available publicly because some building owners are concerned that sharing floor plans will jeopardize security. Some building floor plans are stored by city administrations, where the buildings must conform to city codes and zoning restrictions. Not every city in the U.S. stores building plans. Hesitancy to share geospatial data at this time is international. One French study found that only 15.7% of spatial data in France is open access [12].
Aggregated storage
It makes sense to aggregate indoor map data for the sake of emergency response so that it would be easier for First Responders to access maps in nearby regions that do not happen to be in their jurisdiction. The benefits of aggregating building sensor data for emergency response is well known [19]. Another advantage of aggregated storage is that it would simplify map updating. Buildings change: they might be modified, enlarged, or destroyed by fire. To simplify access for Responders in emergency situations, copies of building maps could be kept in a secured database. Once a map was updated, then all applications that used that map could be notified of the change. Already with the 3D BIM format by IFC, a BIM file update can be sent to all related applications that use a particular file. BIM Web Services can also be used for this purpose.Footnote 21
Storage architecture
One option for geospatial data storage architecture would be a distributed platform to serve and store different geo-data formats, such as the BalticBottomBase project of Gdansk, Poland [8]. In this case, the platform makes the data available in compliance with geographic data sharing standards. Authors Kulesza and Wojcik point out that a file system with geospatial data likely would require external tools, whereas a database could be set up to contain functions that operate on all the spatial data. Another possibility for geospatial storage architecture was presented by [9]. Applications that use indoor maps would still need to allow editing within each application, but widespread indoor map data storage and sharing would represent a convenience to the community of indoor map application developers and users. This explains why, in the paper Introduction, we present an approach to indoor maps that is data centric rather than task (or application) centric
Security
The onus for security and serving the maps only to authenticated users could be part of the storage design or business model. How might this work? In the U.S., indoor map files could be stored within the Homeland Security Information Network (HSIN), that is part of the Department of Homeland Security.Footnote 22 This is already being done for indoor maps of schools and state buildings for West Virginia. Another option is the pay-for-maps model. This is the case with access to a wider map collection from HERE Technologies, owned by several German car companies.
Vision of the database
Creation of a national, interoperable “base map” for the USA recommended at the 2015 Location-Based Services Summit [4] would solve problems of interoperability, access and updating. The United States Geographical Survey (USGS) presently stores geospatial data that is only outdoor.Footnote 23 Adding indoor data to outdoor Geographic Survey (USGS) or to the Homeland Security Network (HSIN) would allow for cross-agency collaboration and improve resources for public safety nationwide. Moreover, adding indoor map data would create opportunities for research and industry to develop tools that use this data. A global vision of an indoor map database would require an enormous volunteer effort to fill in the gaps worldwide. Otherwise, companies such as HERE will charge for map data access.