The evolution of the Global Navigation Satellite System (GNSS) has transformed position-fixing on land, sea and air over the last four decades. The system, which was developed in the 20th century mainly for military purposes, has since gained traction in civil applications of all types, from terrestrial surveying to vehicle navigation and from parcel tracking to precision farming to name but a few. Yet despite the several constellations of GNSS satellites now in operation (GPS, GLONASS, Galileo and BeiDou), none work well, if at all, in areas that lack a line of sight to the satellites.
To meet this need, Indoor positioning systems (IPS) have been developed specifically for the purpose. There is currently no technical standard for IPS and several technologies have been adopted by IPS providers which are widely discussed in the literature.1 These technologies vary depending on hardware required, network availability, and so on. One way of classifying them is to consider the principal means chosen to determine position.2. Based on this, IPS technologies can be classified as those based on Radio Frequency, vision, magnetic field, audio and ultrasound.
Radio Frequency is one of the key technologies widely used for IPS. This is due to the availability of already existing networks and hardware. This category can be further divided depending on the type of the different wireless technologies employed such as cellular, Wi-Fi, ultra-wide band (UWB), Radio-frequency identification (RFID) and Bluetooth.
Cellular positioning has the advantage of widely available signals, and the hardware of customary mobile phones can be used, however, it suffers from low reliability due to varying signal propagation conditions.
Wi-Fi based positioning also has the advantage of existing communication networks which covers a large number of buildings. Furthermore, the availability of Wi-Fi in mobile devices allows Wi-Fi based positioning to be accessible to a large number of users.
However, the Wi-Fi signal also suffers from low signal reliability thus not being precise enough for certain applications. Achieving higher levels of accuracy requires a detailed calibration process and the installation of further access points.
UWB positioning can result in very high positional accuracy even in the presence of obstacles. However, it requires the installation of expensive UWB infrastructure which reduces the interest in the technology for commercial applications.
Bluetooth, on the other hand, has the inherent capability in mobile devices but is expensive compared to the cost of RFID hardware. With the introduction of Bluetooth 4 or Bluetooth Low Energy (BLE), which targets emerging Internet of Thing (IoT) applications, it is increasingly being adopted for indoor positioning as it can also serve as an IoT getaway. Moreover, the technology is already embedded in smartphones as well as being characterised with low energy consumption at both transmitter and receiver ends. The use of Bluetooth as a technology for indoor positioning is further reinforced by the introduction of the BLE beacon protocols, iBeacon and Eddystone, by two of the largest technology companies, Apple and Google respectively.
In plain sight
Vision-based positioning uses the principle of landmarks and maps to determine location. Basic implementation of vision-based positioning is an image captured and compared with a pre-recorded database image which has a position associated with it. Some implementations capture an image of an area and, instead of comparing it to a pre-recorded image, analyse and compare the building structure to a building layout.
In the past, vision-based positioning were mainly of interest to those who worked in robotics as precise locational information is required for applications such as that described in the literature.3
The demand for visual positioning for consumer applications has grown with the evolution of smartphones as these now have the optical and computing capability needed to capture high resolution imagery and handle the image processing algorithms.
The introduction of Google’s Tango device has further sparked the interest of researchers in visual positioning, especially those that incorporate a 3D modelling capability. In the recent years, researchers have also exploited visible light technology, using signals transmitted by LEDs.4
The use of geomagnetic field positioning is based on work similar to that used for outdoor positioning. In 2000, Suksakulchai realised that magnetic field disturbances could be employed for indoor localisation.5 Several papers then emerged describing positioning based on magnetic anomalies. Such methods require no pre-deployed infrastructure but are often combined with other solutions such as Wi-Fi positioning to further improve accuracy.
Other techniques requiring no additional infrastructure are those of audible and ultrasound positioning. Both are based on fingerprinting and generally employ two approaches to estimate a position: passive fingerprinting which uses ambient sound, and active fingerprinting which emits and records specific sound patterns.
Another indoor positioning approach exploits inertial navigation. As this approach uses the accelerometer and gyroscope available in most smartphones, it represents another viable technology which, when used in conjunction with other technologies, finds the initial location and then uses inertial positioning to fine tune the accuracy.
Most of the technologies reviewed above have a similar drawback; they call for a high degree of manual intervention and complex calibration. This makes them expensive and ill-suited to mass deployment on a global scale.
In most cases, further improvement in positional accuracy requires more detailed calibration. This typically entails the deployment of beacons or access points and the creation of a reference point for fingerprinting.
Follow the crowd
Automatic crowdsourcing, pioneered by sensewhere, automates the calibration process by deploying cloud-based algorithms that turn users’ mobile phones and wearable devices into “automatic fingerprinting” engines. These algorithms gather specific information to compile a global and universal database in real-time.
The crowdsourced nature of the algorithms obviates manual surveying, and the database continually improves as data is acquired. The automation of the calibration process also reduces the cost of the initial setup which further increases its attractiveness and commercial potential.
1. M. a. Al-Ammar, S. Alhadhrami, A. Al-Salman, A. Alarifi, H. S. Al-Khalifa, A. Alnafessah, and M. Alsaleh, “Comparative Survey of Indoor Positioning Technologies, Techniques, and Algorithms,” in 2014 International Conference on Cyberworlds, 2014, pp. 245–252.
2. Y. Gu, A. Lo, and I. Niemegeers, “A survey of indoor positioning systems for wireless personal networks,” IEEE Commun. Surv. Tutorials, vol. 11, no. 1, pp. 13–32, 2009.
3. E. Rivlin, I. Shimshoni, and E. Smolyar, “Image-based robot navigation in unknown indoor environments,” in Proceedings 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2003) (Cat. No.03CH37453), 2003, vol. 3, pp. 2736–2742.
4. J. Armstrong, Y. Sekercioglu, and A. Neild, “Visible light positioning: a roadmap for international standardization,” IEEE Commun. Mag., vol. 51, no. 12, pp. 68–73, Dec. 2013.
5. B. Li, T. Gallagher, A. G. Dempster, and C. Rizos, “How feasible is the use of magnetic field alone for indoor positioning?,” in 2012 International Conference on Indoor Positioning and Indoor Navigation (IPIN), 2012, pp. 1–9.
Prof. Tughrul Arslan is a co-founder and CTO of Edinburgh-based indoor location provider sensewhere Ltd. (www.sensewhere.com) as well as Chief Scientist at Sofant Technologies, a provider of smart antenna and tunable RF solutions, also based in Edinburgh. He holds the Chair of System Level Integration at the University of Edinburgh and is also a consulting scientist to NASA Jet Propulsion Lab in Pasadena, USA. He previously acted as the Director of the Cardiff Advanced IC Prototype Centre.