FM radio is another possible candidate, utilizing frequency‐division multiple access (FDMA) to split the wireless band into a number of separate frequency channels that are used by stations. FM band ranges and channel separation distances vary in different regions, but the pervasiveness of FM signals can enable their use for indoor localization. Typically, radio waves operating in the frequency band 87.5 to 108.0 MHz are part of the FM spectrum. Due to the passive nature of the client devices, FM can be used in sensitive areas where other RF technologies are prohibited for safety or security reasons. Unfortunately, FM signals lack timing information, which limits their use in certain localization techniques (such as the time‐based trilateration techniques discussed in Section 37.5.1.2).
37.4.2.3 Challenges
In general, the propagation of RF signals in indoor environments faces several challenges. Certain materials in the indoor environment affect the propagation of RF waves. For example, materials such as wood or concrete attenuate RF signals, while materials such as metals or water cause reflections, scattering, and diffraction of RF waves. These effects lead to multipath radio wave propagation, which prevents accurate calculation of the distance between the transmitter and the receiver in indoor environments. The propagation of RF waves is also affected by changes to the physical indoor environment (e.g. movement of people, rearrangement of furniture, structural modifications). In these environments, the RF properties are highly dynamic, and a radio map captured at a certain point in time cannot be used reliably for localization without accounting for these dynamic changes [12]. Moreover, while some solutions operate within a reserved radio band [18], most solutions utilize open spectrum bands. This means that these solutions must account for the increased risk of interference due to other systems sharing the same frequency bands of the radio spectrum. The usage of radio transmitting devices is also restricted in some cases, for example, in critical areas of most healthcare facilities, according to recommendations made by the Association for the Advancement of Medical Instrumentation (AAMI) [20] and other standards or regulatory bodies. Such restrictions limit the deployment of localization systems based on non‐broadcast RF waves.
37.4.3 Ultrasound
Sonic waves are mechanical vibrations transmitted over a solid, liquid, or gaseous medium. The sonic waves produced by vibrations below and above the threshold of human hearing are known as infrasonic and ultrasonic waves, respectively. There are indoor localization solutions that propose the use of ultrasonic range finders and sonars [21]. The relative distance between two devices can be estimated from ToA measurements (see Section 37.5.1.2) of ultrasound pulses from the emitter to the receiver, and thus ultrasound signals can be used to estimate the position of the emitter tags from the receivers. Typical ultrasound systems operate in the low‐frequency band compared to the other RF signaling technologies. In contrast to RF waves, the ultrasound ToA operating range is 10 m or less due to the specific decay profile of the airborne acoustic channel. Doubling the distance causes the signal’s sound pressure level to attenuate by 6 dB due to radial intensity attenuation and absorption, which translates to an inverse quadratic attenuation in 3D space. In general, ultrasound signals are unable to penetrate walls, and they reflect off most indoor obstructions (furniture, people), resulting in echoes that can lead to localization inaccuracies. It has also been observed that high levels of ambient noise prevent accurate detection of the sonic signal; co‐interference caused by the presence of multiple sonic emitters in the environment also leads to errors. Variations in the speed of sound over air are another challenge: for instance, temperature variations are known to affect the speed of sound in air [22]. Therefore, sonic‐wave‐based systems cannot be used in environments with frequent and drastic temperature changes.
37.4.4 Inertial and Mechanical
Whenever energy is exerted due to the mechanical movement of a moving subject, the energy can be measured and used for localization in indoor environments. As an example, a “Smart Floor” [23] was proposed with metallic plates that were instrumented with load cells, which used mechanical coupling between a moving person and the load cells. The plates were laid on the floor, and the signal captured via the load cell was processed in order to identify the person walking over a plate and the path they were taking. A more common example of exploiting mechanical energy for localization is via accelerometer (to measure acceleration) and gyroscope (to measure angular rotation) sensors. Such “inertial” sensors that are part of IMUs commonly found in smartphones can be used to estimate the trajectory of motion for a moving person or object, which can help with their localization in indoor environments [24]. In particular, such sensors are very useful to estimate the stride length and step counts for a person in motion, to determine their displacement over time. Techniques that use inertial sensors for localization are often referred to as “dead reckoning” techniques, as the location estimates provided by the sensors depend on previous measurements to estimate the absolute position or orientation of the object being tracked at any given instant. A challenge with using inertial measurements for localization is that the inertial sensors are susceptible to drift due to thermal changes in the circuitry of the sensors, calibration issues, and inherent noise [25].
37.4.5 Other Signals
There are a few other signals that can aid with indoor localization. Atmospheric pressure can be captured using barometric/altitude/pressure sensors, and used to provide estimates of the altitude of the person or object to be tracked. Magnetic readings captured by a (digital) compass can also be used for heading (direction) estimation. Most IMUs today include three perpendicular magnetometer sensors to measure the strength and/or direction of a magnetic field, along with traditional 3‐axis accelerometer sensors for motion estimation, and 3‐axis gyroscope sensors to measure angular rotation. However, spurious electromagnetic field disturbances can affect the readings of the magnetometer sensors when in proximity to metallic structures or radio‐wave‐emitting devices.
In general, the signals discussed above can help improve the accuracy of indoor localization when used in tandem with other more robust and comprehensive localization signals, for example, dead reckoning or RF‐signal‐based localization.
37.5 Indoor Localization Techniques
Having identified the commonly used signals for indoor localization, we now present a survey of various indoor localization techniques that have been proposed and evaluated to date. We classify these techniques in this section based on the measuring principles used: triangulation, fingerprinting, proximity, dead reckoning, map matching, and hybrid techniques. Typically, for all of these different types of techniques, there are two main approaches for deployment: (i) developing a custom signaling and network infrastructure, and (ii) reusing an existing network infrastructure (e.g. existing Wi‐Fi APs in a building). With the first approach, it is possible to control the physical specification and, consequently, the quality of the location sensing results; whereas the second approach has much lower costs as it avoids expensive and time‐consuming deployment of infrastructure [26].
37.5.1 Triangulation
Triangulation is a family of wireless radio‐signal‐based methods that use the geometric properties of triangles to determine location. The methods can be broadly classified as angulation‐based and lateration‐based [27]. Angulation locates an object by computing its angles relative to multiple