Taking High Precision Positioning Indoors
October 15, 2021
Bluetooth currently offers a new answer to the high precision indoor positioning conundrum. We tested the technology in an industrial warehouse proof of concept trial.
Location awareness: It’s one of the key benefits that the IoT technology has brought businesses – and consumers. For a modest price, fleet managers can track their vehicles, logistics firms their shipped goods, and farmers their livestock – all in real time. Meanwhile, you and I could purchase a connected device to keep an eye on an aging relative, our pets, or valuable belongings such as a car.
Enabled by global navigation satellite systems (GNSS), including the GPS, GLONASS, BeiDou, and Galileo, location technology has found its way into almost all aspects of our economy, and our daily lives. All the while, GNSS technology has improved year after year, with achievable accuracies tightening from meters to mere centimeters, time to first fix moving from tens of seconds to seconds, and service availability expanding its footprint to cover even the densest urban locations.
Yet, still today, high precision positioning has one major blind spot: the great indoors. Weak GNSS signals fail to penetrate effectively into most indoor spaces. As a result, the improvements in efficiency enabled by continuous location awareness have been withheld from countless activities:
- Hospital managers tracking medical devices, patients, and staff.
- Airport operators optimizing luggage handling and quickly locating late passengers.
- Manufacturing companies automating production processes.
- Service and retail organizations tracking customer behavior and backend operations.
- Warehouse managers upgrading their operations with unmanned ground robots.
With satellite signals largely unavailable in these settings, a series of other technologies have been suggested to fill the void. Devices equipped with cellular modems can, for example, use cellular signals to achieve a position estimate relative to nearby cell towers using so-called network fingerprinting or more sophisticated time-of-flight techniques. Wi-Fi-equipped devices can use similar approaches to locate themselves relative to Wi-Fi hotspots. And those featuring Bluetooth can use the received signal strength indicator (RSSI) to derive an approximate distance estimate relative to deployed Bluetooth beacons.
*Approximate figures. The exact values depend on the details of each deployment.
These technologies, however, all suffer from limitations that have restricted their uptake in the types of use cases listed above, where the accuracy, availability, ease of use, and affordability of GNSS technology have set an ambitious benchmark. Cellular and Wi-Fi based location technologies require relatively high hardware costs to achieve accuracies that fall short of expectations. And despite its low accuracy, Bluetooth RSSI has managed to get a foothold in applications requiring room-level location accuracy, due to its low cost, low power demand, and compatibility with the majority of connected devices in circulation.
In 2019, the Bluetooth SIG leveled up its game in the indoor positioning arena with the introduction of Bluetooth Direction Finding. The approach, which uses a new type of Bluetooth signal and multiantenna arrays to measure the angle of travel of a Bluetooth message between a mobile tag and one or several static anchor points, brings a new potential answer to the indoor positioning conundrum that, arguably for the first time, checks all the boxes. High precision? Check. Ease of deployment? Check. Low device cost? Check. Low power demand? Check.
Since they were first introduced, Bluetooth-based indoor positioning solutions have garnered considerable interest, with ABI Research predicting a 28.3% CAGR for Bluetooth tag shipments from 2019 to 2025, with the largest increase by far (64.2%) expected in smart offices and the largest absolute number (over 163 million) in the warehouse and logistics vertical. Because of its complementarity with our outdoor GNSS solutions, the vast global ecosystem of businesses developing solutions based on the technology, its low power and cost, and the submeter positioning accuracies that it can deliver, we at u-blox have focused considerable R&D activities on promoting the uptake of the technology.
How Bluetooth Delivers High Precision Indoor Positioning
The underlying technology for Bluetooth indoor positioning is Bluetooth Direction Finding. As indicated by its name, Bluetooth direction finding, which comes in two flavors, makes it possible to determine the direction of travel of a Bluetooth signal between a mobile tag and a fixed anchor point. In the case of angle of arrival (AoA), the anchor point computes the direction of the incoming signal that was transmitted by the tag. In the case of angle of departure (AoD) the roles are flipped, and the tag is tasked with calculating the angle at which the signal was transmitted from the anchor point.
In this article we focus on AoA, which is the better fit for indoor positioning solutions, while AoD has advantages for indoor navigation solutions.
To evaluate the technology, we built an AoA-based Bluetooth direction finding demo in our offices in Malmö, Sweden, in which we programmed a servo mounted on a direction-finding anchor point to track a moving Bluetooth tag based on the angle of arrival output calculated in real time.
There are two secrets to how Bluetooth direction finding works. The first is a new Bluetooth direction finding signal featuring additional data called a constant tone extension (CTE). Whereas the remainder of the Bluetooth message is modulated to carry data, the CTE consists only of a string of “ones.” As a result, the receiver can use this part of the message to accurately measure phase differences between signals. Which brings us to the second secret to how Bluetooth direction finding works. Inside each anchor point hides not a single antenna but a multiantenna array.
The figure above illustrates how the direction finding signal emitted by the mobile tag reaches the individual antennas of the static anchor point’s antenna array. Because of the difference in distance traveled, each antenna receives the signal with a slight phase shift relative to the others, which is measurable thanks to the CTE. Algorithms running on an MCU embedded in the anchor point can then parse this data to calculate the signal’s angle of arrival with an approximate accuracy of +/- X degrees.
When instead of a single anchor point multiple anchor points are used, the angles of arrival from several anchor points can be used to triangulate the approximate location of the tracker. This requires entering the precise positions and orientations of the anchor points into the positioning engine, which then runs another algorithm to compute the location – in 2D or in 3D – of the tagged asset based on the angles of arrival calculated by each anchor point.
In a simple 8-by-6-meter office setting with four anchor points mounted in the corners, we achieved an average accuracy of under 1 meter with 95% probability.
Trialing the Technology in an Industrial Warehouse
We trialed our Bluetooth indoor positioning solution in a real-world industrial warehouse – a typical deployment scenario for asset tracking applications. The 30-by-50-meter warehouse had metal shelves to store equipment and boxes. While the Bluetooth specification defines the lower layers to handle the raw RF data, it doesn’t specify the algorithm to calculate the actual angle of arrival. For the trial, we developed an efficient algorithm to run on the embedded MCU in the Bluetooth chip while still providing both high accuracy and update rate. In particular, we optimized the RF-front-end, the antennas, the embedded algorithms running in the anchor points’ Bluetooth modules, and the wireless connectivity backbone to connect the anchor points into a network.
A Bluetooth tag and the L-shaped antenna array used in our anchor points are depicted in the image below.
For our trial, we used ten anchor points to cover a six-meter-high volume with a roughly 1000-square meter footprint. After careful planning and preparation, the installation of the positioning system was non-disruptive, taking only around two hours. To maximize line of sight between the tracker tags and the multiantenna arrays, we mounted the anchor points between three to five meters off the ground.
We simplified deployment by using third-party tracking software, in our case we used Traxmate, which let us easily enter the positions and orientations of the anchor points and configure the positioning engine using an integrated API. Finally, we set up Wi-Fi communications backbone between each anchor point and the positioning engine.
In our trial, we placed extra effort into designing the setup to deliver reliable performance in an indoor environment that captured much of the complexity that most indoor deployments will have. First, we strategically placed the anchor points to maximize the probability of line of sight between all likely tag positions and at least three anchor points. On top of that, we had to deal with multipath effects, caused, for example, when radio signals bounce of walls. The algorithms running in our anchor points to calculate angles include multipath mitigation, delivering robust performance, even in our warehouse’s challenging radio environment.
Our experience carrying out this proof-of-concept deployment has only strengthened our conviction that Bluetooth high precision indoor positioning can deliver on its promise. For one, if well deployed, it can deliver the sub-meter accuracies expected by emerging indoor positioning use cases. As is typical for Bluetooth devices, the cost of the required hardware is well below that of competing technologies, as are the power requirements. Deploying indoor positioning solutions has long been challenging. Integration of the hardware setup we used with the web interfaces, such as the one developed by Traxmate, have done a lot to simplify deployment.
With the release of Bluetooth direction finding, the Bluetooth SIG has offered a strongly compelling solution to the indoor positioning conundrum that addresses many of the shortcomings of solutions that are currently on the market. Our experiences with the technology in a “follow-me” use case, in which direction finding was used to steer a servo (which could, for example, hold a camera), in an office-scale indoor positioning trail, and a real-world industrial warehouse proof-of-concept all underscored the technologies potential to transform indoor use cases with location awareness just as GNSS technologies have done outdoors.
To learn more about the hardware we used in our trials and proof of concept and to benefit from the algorithms we developed to compute the angles of arrival at each anchor point the estimate position in the positioning engine, and to mitigate multipath effects to increase the reliability of the solution in real-world deployments, we encourage you to take a look at our Bluetooth Direction Finding and Bluetooth indoor positioning explorer kits and to get in touch at https://www.u-blox.com/en/contact-u-blox.
Erik Carlberg is a senior product manager in the Short Range Radio product center within u-blox AG. He is responsible for the roadmap of Bluetooth and Wi-Fi modules, targeting the industrial market.
Erik has 20 years of experience from the wireless industry. He holds a Master of Science degree in Electrical Engineering and a Bachelor in Business Administration, both from Lund University in Sweden.