When we first think about positioning or localization, GPS inevitably comes to our mind. We have become so dependent on mapping services that one can barely fathom other uses for localization technologies. We use navigation services in our everyday lives to get a route from point A to point B, to explore a city, or find out where an unfamiliar place is. However, you might be surprised to find out that there are plenty of other localization technologies (even some based on the smell!) and applications beyond the good old GPS.
First, let’s see how and where we can use localization technologies.
We are sometimes too invested in our own lives to think about all the hard work and effort that is made to create objects and services that we use daily. However, you should know that a great deal of localization goes into manufacturing and automotive delivering services. Sending supplies to people’s homes and delivering healthcare kits to hospitals became especially helpful during the pandemic times, and it is said to be “the new normal”.
Delivery services generate a revenue of US$200 billion in the United States and (semi-) autonomous drone delivery is expected to completely transform this industry by reducing traffic load and delivery times, increasing road safety, and improving accessibility. High accuracy positioning systems are essential to enable the drones’ autonomy and will play a crucial role in this industry.
While the concept of smart homes might seem futuristic for some of you, we might soon view a home that reacts to where we are as indispensable as our microwave today. When we say “home” what we actually mean is the collection of interconnected electronics, appliances, and gadgets in our home that work together for our benefit. For instance, a TV show that you were watching in the living room could be offloaded to your phone when you go to the kitchen or to your computer when you are in front of it. The central heating system could adjust the temperature in the entire house based on the room in which you are at a given moment, thus reducing both your heating bill and your carbon footprint.
These technologies might seem like whims, but they can significantly improve the quality of life of people with disabilities or who need assisted living. Therefore, this has led to the concept of Ambient Assisted Living, an ecosystem of sensors, devices, and wireless networks that monitor our health and well-being. Localization systems for ambient assisted living can ensure the safety of patients suffering from Alzheimer’s disease and restore the independence of the elderly.
Speaking of healthcare, wearables that measure our heart rate, count the number of steps we take each day, and report how many calories we burn are already ubiquitous. Devices equipped with localization services are especially helpful for athletes, as they help them create a running route, monitor their pace and other general parameters to stay on track.
Now that we know what to use localization for and what a localization system is supposed to do, let’s see how we can actually implement a localization system. There are many ways to classify a localization system, for instance based on their accuracy, environment in which they operate, or energy consumption. Here, however, we will split localization systems based on whether they use radio frequency (RF) signals (see an example in the figure below) or not. Radio frequencies are roughly situated between audio and infrared frequencies and have the advantage that they can usually penetrate objects. On the other hand, some non-RF technologies based on light have localization accuracy in the range of millimeters, so we might prefer them when we can ensure line-of-sight (LOS) conditions.
Some of the most popular RF localization technologies are Bluetooth, Wi-Fi, and Ultra-Wideband. In the figure above, you will find an example of the waveforms. As you may see, those signals can have different shapes or forms.
Bluetooth Low Energy (BLE) is a special case of Bluetooth technology designed for low power communication. In positioning, this technology is mainly used for proximity detection. We can compute the distance between two devices based on the strength with which the signal arrives at the receiver. Ideally, the signal strength should decay linearly with the square of the distance. In reality, it’s more complicated than this because the signal can reflect on the surrounding objects and then add up – constructively or destructively – at a receiver in a way that’s not necessarily proportional to the distance. A Bluetooth sensor will therefore transmit signals (also called beacons) every second or minute to sense if there are any other Bluetooth devices in its proximity. If we know the locations of some reference devices called anchors (we need at least three for 2D positioning), we can compute the location of the user at the intersection of circles centered at the anchors and with a radius equal to the distance between each anchor and the user, like in the figure below. This method is known as trilateration and can be used with any technology which can estimate the distance between two devices.
Wireless Local Area networks (WLANs) enable data transfer at high speeds. Wi-Fi (you must have heard about it more often) is a brand name which marks some of the WLANs. With this technology we can perform another type of localization, using “fingerprints” of the signal. Remember when we said that the signal can reflect on the surrounding objects and add up in different ways at the receiver? This imprints upon the signal certain characteristics – maybe a particular received signal strength or a channel impulse response – depending on the location of the receiver, which are known as fingerprints. If we have a database of signal fingerprints at different locations in a room, we can compare the characteristics of the signal at a certain moment with the stored fingerprints and see which one resembles it more closely. The location corresponding to the closest match will be the location of our user. Localization via fingerprinting is not a typical case, as the technology was not specifically designed for this purpose. In this scenario, where the technology was not initially designed to be used in localization, the signals are called as signals of opportunity.
Ultra-wideband (UWB) is another technology that can perform positioning. As their name implies, UWB signals have a very large bandwidth (over 500 MHz), which means that they also have a very high time resolution (their pulses are very short). Thanks to this property, they are suitable for time-based localization methods, in which we compute the user’s location based on the time of arrival of signals. For instance, if we know the time of flight of the signal and consider that the signal travels at the speed of light, we can compute the distance between two devices. If we know the distance between two devices, we can go back to the trilateration method. UWB provides sub-meter accuracy in ideal scenarios, which is a huge advantage for asset-tracking tasks.
Non-RF localization systems are perhaps even more diverse than RF ones. Sound can be used for localization purposes and performs quite well even with obstacles, such as an opaque wall (see these studies for more details). Light-sensing techniques are used to localize the object in the space and claim to provide mm-level accuracy with low-power consumption. Companies such as Wearnotch provide systems with motion capture sensors (pictured above) and even make it possible to visualize the recorded data into 3D models. Special algorithms are designed to implement the odor-based localization. Humans are able to sense only 10 categories of odor. Inspired by animals and their outstanding ability to smell, researchers are working to create a robot that will be endowed with this skill. There are three steps to follow: determine the presence of a chemical in space, find the source based on chemical or another sense, and complete it with odor recognition. Although there are some sensors already available called portable “digital noses”, the technology is still at the research stage and isn’t yet ready for a mass adoption.
Now let’s see how these technologies compare in terms of the most important characteristics of a localization system: accuracy and precision, privacy, energy efficiency, ease of deployment, and user scalability. We’ll first introduce these notions and rate the most important localization technologies (GPS, BLE, Wi-Fi, UWB, and motion capture systems) on a scale from 1 to 5 according to how well they perform in each of these areas.
Accuracy and precision. First of all, any localization system should be accurate and precise. These terms are not mutually exclusive as some might believe. The figure above shows an example of accurate and precise location estimates. Accuracy in positioning stands for the closeness of the experimental measurements to the exact values, coordinates on the map, whereas precision relates to the “exactness” of the measurements. The exact accuracy and precision depend on the use case. For instance, in global positioning we don’t need centimeter-level accuracy and we can usually tolerate errors of several meters (although it might be very annoying when your navigator does not correctly detect the street side you are on and you have to cross the street to reach your destination—we know, first-world problems). If we’re talking about remote surgery, on the other hand, an error of 1 cm in locating the scalpel can perhaps leave you without a much-need artery.
The most accurate and precise localization systems are without doubt motion capture systems, which can localize a target with sub-millimeter errors. UWB also performs quite well in this area and usually has an accuracy and precision under 15 cm. BLE and Wi-Fi with fingerprinting, angle of arrival, or time-of-arrival have at best decimeter-level accuracy, but most likely in the range of 1–2 m. Finally, consumer-grade GPS is the least accurate and precise of all since its positions usually have errors in the range of meters.
Privacy. Our smartphones continuously keep records of the places we visit and know the approximate locations of our homes and work offices (in case the location-tracking engine is on). Location information is collected via triangulation and cell towers. So if you are concerned about this you are not alone. Location privacy refers to the possibility to determine one’s identity and track their location, so this term stands for the right of individuals to decide on how, when, and to which extent their location information is collected, processed, and shared with third parties. Privacy matters, maybe way more than you could imagine. Using a leaked location data set, journalists were able to recover the trajectory of the President of the United States (one of the most safely-guarded persons on the planet) and find the home and work places of others.
The most privacy-friendly technologies are the ones where the user does not need to transmit any data to the location server and can compute its position locally. This is the case for RSS-based BLE, GPS, motion capture systems and some time-based localization methods with UWB signals. On the other hand, Wi-Fi fingerprinting is privacy-sensitive because RF fingerprints need to be sent to a database to find a match.
Energy efficiency. High-end motion capture systems usually rely on cameras and other energy-hungry devices to locate users. At the other extreme, BLE and some types of UWB devices have very low energy consumption suitable even for small wearables. Wi-Fi and GPS are somewhere in the middle—good enough for smartphones or some smartwatches, but not efficient enough for highly energy-constrained devices.
Ease of deployment. Most localization systems need some sort of infrastructure to be able to localize a target. In the case of GPS, this is represented by the satellites. Although they are difficult to deploy, satellite networks are already in place and any user equipped with a GPS receiver is free to use them, so building localization systems based on GPS signals is ultimately easy to do. BLE and Wi-Fi transceivers are, in many instances, already installed in buildings so we can build localization systems on top of this infrastructure. However, fingerprinting-based Wi-Fi localization systems require a training step which is, in most cases, quite difficult to perform and prone to changes in the environment. Most public spaces do not already have a network of UWB devices in place for localization, so its initial cost is higher than that of BLE or Wi-Fi localization systems. Once deployed, however, it does not require any training step and is ready to use. Motion capture systems are the most difficult to deploy. They require cameras or other light-emitting devices which must have line-of-sight to the target for the localization to take place.
User scalability. In many cases, a localization system needs to provide locations to a very large number of users. The systems which perform the best from this point of view are usually those where the user does not need to transmit data but can localize itself based on messages from satellites or anchors. GPS is obviously able to satisfy this demand for billions of users around the globe. Localization based on the RSS from BLE beacons can also scale with a large number of users. Some time-based UWB methods (for instance, using the time-difference of arrival of signals) can also accommodate many users. In the case of Wi-Fi fingerprinting, since RF fingerprints need to be sent to a server, it might not be able to satisfy an extremely large number of requests. Finally, many motion capture systems actually do not require the user to send any uplink messages but given the fact that they can only cover a small area (usually less than 10 x 10 m) and the users need to be in line of sight with the anchors, in reality it would probably not be able to accommodate tens let alone hundreds of users.
The figure below summarizes our discussion and rates each localization technology in all criteria.
We hope that through this article you gained a brief understanding of localization systems, perhaps found some ways to bring them in your everyday life, and learned about which angles to consider if you decide to give them a try.
by Laura Fluerătoru and Viktoriia Shubina