Bluetooth Low Energy: the elusive promise of the perfect locating technology
The path to a great Bluetooth-based locating solution is strewn with broken promises and blind alleys. Where do we go from here?
Bluetooth Low Energy (BLE) devices are inexpensive, ubiquitous, and highly configurable. Compared with real-time location systems (RTLS) that use other technologies, BLE-based systems use inexpensive hardware and open standards, with most of differentiation in the software layer that refines the signal into valuable location data. BLE powers many of the location devices consumers interact with (Tile, Airtags) but it still struggles to achieve the level of accuracy and responsiveness needed to support the spectrum of clinical use cases.
In college I spent a semester studying in Barcelona. We had an art history professor, who, armed with a gigantic mustache and a slide projector showed us photographs of Moorish architecture like The Alhambra or Grande Mosque. He kept repeating that these were prime examples of “rico diseño hecho con materiales pobres”, which generally translates to “fine design made from cheap materials”. The gist of what he was so impressed with, and wanted us to be impressed with was how something of high value could be created from something of low value through great craftsmanship. The same concept applies to BLE real-time location systems. When designed well, the systems leverage BLE hardware (cheap components) to create high value (accurate location data) through craftsmanship (machine learning algorithm and sensor fusion techniques).
Most RTLS that use other technologies to determine locations do not share this approach. In contrast, they use hardware that is more expensive, but resolving that data into an accurate location requires much less processing power and time to achieve a high level of accuracy. This is true for RTLS that use infrared (IR) light or ultrasound signal. Because the reach of each of these signal types is constrained by walls, it does not take much compute power to determine whether a tag’s location is inside a room or in the adjacent hall. Of course, using these types of signal technologies has drawbacks as well, such as hardware cost, system complexity, proprietary technology, and restrictions on current or future interoperability. With the difficult decisions posed by these factors, many hospitals have chosen not to adopt a locating technology, yet, as they have been waiting for the “right time”. Many solution seekers believe that the “right time” is when a BLE-based solution becomes available that can meet the high demands of accuracy and responsiveness needed for many clinical use cases.
With this growing pent-up demand, the arrival of Bluetooth Low Energy (BLE) in 2011 brought a wave of optimism about the potential for BLE to crack the nut of a low-cost do-all system.
The fundamental challenge in creating a BLE real-time location system has to do with balancing trade offs. Usually fully satisfying one dimension of need (accuracy, timeliness, battery life) comes with unacceptable effects on the other dimensions. So many original equipment manufacturers (OEMs) generally have to make hard decisions about a balance that they believe will be most valuable to their target markets. Many systems have configurations that provide a limited range of flexibility, but the trade offs still exist.
Here are some examples:
If you want a high level of responsiveness (low latency), such that a location change is recorded quickly when, this has the potential of reducing battery life for both tags and infrastructure to an unacceptable duration.
If you want a high level of location accuracy, it could require the collection of a significant volume of data, sending it to the cloud, and processing it through an algorithm to achieve a high confidence level. In ter, this results in higher requirements of time, power, and cost to collect, transmit, and process the data.
In addition to these challenges, BLE can struggle at the most fundamental level: obtaining quality raw data. Because BLE exists in a crowded range of the radio spectrum (2.4 GHZ), the raw data collected about signal strength can be degraded, setting the stage for diminished location accuracy.
Some factors which can affect BLE signals are:
Signal attenuation (BLE signals can be significantly attenuated by walls, metal objects, people, etc. This makes it hard to maintain a consistent signal strength that correlates well with exact distance from the receiver.
Multipath propagation: BLE signals often reach the receiver via multiple paths (reflecting off surfaces). This causes unpredictable signal distortions that don't provide a reliable indicator of straight-line distance.
Environmental dynamics: Movement of people and objects within a building causes constant fluctuations in signal propagation which reduces location accuracy.
Interference: There may be interference from WiFi, other BLE transmitters, etc. This can skew distance and direction calculations.
But The Alhambra and the Grand Mosque are not great works because plaster was cheap. They are impressive works because tremendous value has been added to a cheap material through craft. In the case of Bluetooth Low Energy, the value add needs to come in the form of advanced algorithms that have the capability of resolving signals into locations with a very, very high level of accuracy. Of course these may happen in conjunction with other advances, such as in battery capacity, or decreased power consumption of radio devices.
Engineers are working on all dimensions of this challenges, but I believe that it's proving to be a more difficult (and more expensive) nut to crack than many hoped for a few years ago. It's only recently that we have achieved a coherent convergence of a low-cost, low-compute and machine learning enabled environment that has allowed a level of performance that covers the majority of use cases in a clinical setting.
The highway to a full-featured BLE-based RTLS is strewn with broken promises — one does not need to look hard to find disappointed customers or failed proof of concept deployments. However, I think we are getting close to achieving a semblance of balance between the critical dimensions of cost, accuracy, responsiveness, and ease of use that will put this technology within the reach of many organizations.
Or, a new technology may burst onto the scene and leapfrog all of this. We shall see.