3 keys to enterprise-grade Bluetooth beacons

Mist Systems combines cloud intelligence with on-premise access points to deliver accurate indoor location services

3 keys to enterprise-grade Bluetooth beacons
Credit: IDG Communications, Inc. Photo contributed by Matthew Mikaelian.

The world is at an inflection point, where smart devices (phones, tablets, and laptops) are taking over as the predominant computing platform. While this can present various challenges from a wireless operations and management standpoint, it also opens up huge opportunities for businesses. That is because smart devices are location-aware, and location is a strategic attribute for engaging with customers or employees and tracking high-value mobile resources. 

Thanks to iBeacon and Eddystone, the Bluetooth Low Energy (BLE) beacon technologies used by Apple and Android respectively, the vast majority of mobile computing platforms are location ready. The challenge to capitalizing on location awareness comes on the infrastructure side. How can the wireless network take advantage of BLE in a scalable, reliable, and cost-effective manner?

Deploying and managing physical beacons for BLE can be both expensive and time-consuming. Installation often requires comprehensive site surveys, which have to be redone whenever beacon placement changes (or the RF environment is altered). In addition, batteries in the beacons are expensive to replace, and the beacons themselves can be lost or stolen, creating more headaches. Furthermore, wireless networks generally lack the scale to handle a large number of BLE devices, complicating performance and management of BLE networks.

Mist Systems overcomes these challenges of physical beacons by integrating BLE location services into its enterprise wireless networking platform. How does it work? Mist’s Intelligent Wireless Cloud (IWC) platform combines three key ingredients to deliver flexible, scalable, enterprise-grade location services: virtual beacon technology, machine learning, and wireless access points with directional antennas. Let’s take a closer look.

Wireless APs with directional BLE antennas

The first step in a BLE location service is to blanket the room with BLE signals. This is best achieved via directional antennas powered by a single Bluetooth transmitter sending unique RF energy in multiple different directions. This directionality results in much higher accuracy than traditional environments using physical beacons. Let me explain why.

Imagine being in a room with a single lightbulb. If you could measure how strong the light is, you could determine how far you are from its source. The closer you are, the brighter the light; the farther away you are, the dimmer it appears.

Typical physical BLE beacons operate by sending RF energy in all directions like a lightbulb. Mobile devices simply detect the strength of signals from the BLE beacons and tell the mobile app which one is the brightest.

If there were several lightbulbs in different locations (imagine different colors for identification), it would be possible to triangulate among them to determine your location on the floor plan. This technique, which is used by most BLE systems, requires multiple BLE tags in different locations to work with any degree of accuracy.

Now imagine that instead of a lightbulb spreading light in all directions, we use a flashlight shining in one direction. In this case, most of the light shines directly in front of the flashlight, with the beam fading in brightness the farther from the center you get. The strength of the light now provides an accurate indication of where you are. The beam is brightest in a line directly to the device, but fades to either side (see Figure 1).

mist cloud Mist Systems

Figure 1

Directional BLE antennas in access points create a beam like that of a flashlight, with more energy pushed in front of the directional antenna than out the back or to the sides. The energy forms a power distribution much like an ellipse. A probability weight is then assigned to each point in the location map. The bigger the gap between the measured signal strength and the expected signal strength, the lower the probability the device is at that location. By combining and then analyzing probability surfaces for every directional beam, the most likely location of a device is determined with exceptional accuracy (see Figure 2).

mist sum of surfaces 2 Mist Systems

Figure 2

Machine learning in the cloud

The Mist system uses machine learning in the cloud to fine-tune location accuracy and to offload location computations from the devices themselves to save battery life.

Machine learning promises to solve complex problems that can’t easily be described with a simple set of rules. There are two types of machine learning: supervised and unsupervised. In supervised machine learning, a person examines the answers produced by a computer and indicates whether or not the answer is correct. The computer algorithms adapt to changes in the data based on the human input. In unsupervised machine learning, only data is entered into the system. The algorithms still adapt their conclusions as the data changes, but these adaptations are based on the incoming data alone.

Determining the characteristics of an RF environment is a perfect example of a complex problem that can benefit from machine learning. That is because adding people or furniture to a room changes the RF characteristics of the environment, so it's difficult to predict RF behavior using static rules. And if a BLE solution cannot predict RF behavior, then it cannot pinpoint the locations of mobile users.

To address this challenge, Mist uses unsupervised machine learning. The system analyzes the Received Signal Strength (RSSI) observations gathered from iPhones, Android smartphones, and other mobile devices as they come and go, and continuously updates the RF model for each device type. This continuous analysis ensures that the RF model adapts to changes in device types and to changes in the wireless environment, without the need for site surveys and manual calibration.

mist ble machine learning Mist Systems

Figure 3

Mist Systems’ machine learning takes location estimates from everyday use, examines them, detects the RF characteristics based on the input, and adapts the path loss formula for each device in the environment (see Figure 3). The goal is to seek maximum agreement among the results of many location estimates, then construct individualized path loss formulas based on that data. Using the device-specific path loss formula, Mist can determine the distance between the device and each AP in the environment, as well as triangulate the user’s position on the floor with high accuracy.

Virtual beacons

The cost and effort associated with deploying and maintaining physical beacons for BLE has prevented many sites from embracing location services entirely. Physical beacons don’t scale, which has prevented BLE from becoming a mass-market technology to date.

New virtual beacon solutions solve these issues. By moving the beacon functionality into existing APs and enabling beacon messages (and zone analytics) to be configured anywhere on a floor plan via software (see Figure 4), virtual beacons create a much better operational experience than physical beacons. An unlimited number of virtual beacons can be deployed in a physical environment with the simple click of a mouse on a floor plan (or programmed via APIs).

Virtual beacons offer many advantages:

  • No batteries
  • Beacons are easily deployed, configured, and moved via software (no on-site visits required)
  • No risk of loss or theft or movement from a beacon’s original position
  • Building aesthetics are not affected by the deployment of numerous physical beacons
  • Virtual beacons are “stackable,” meaning that different applications and tenants can get different messages
  • No site surveys or ongoing calibration required
  • Virtual beacons can co-exist with existing BLE deployments, so you can work with your existing investment
mist virtual beacons Mist Systems

Figure 4

Location inside

Intel Centrino changed the wireless world forever in 2003 by bringing 802.11 to the masses. The wireless world is at a similar tipping point right now. Every smartphone, tablet, and laptop has BLE, making these devices ready for new indoor location-based services.

With the advent of modern wireless platforms such as the Mist Intelligent Wireless Cloud, combining enterprise-grade BLE access points, machine learning, and virtual beacon technology, the infrastructure is now ready to support the demand.

Given the breadth of use cases, which stretch across almost every industry, BLE location-services promise to usher in a new era of smart mobility.

New Tech Forum provides a venue to explore and discuss emerging enterprise technology in unprecedented depth and breadth. The selection is subjective, based on our pick of the technologies we believe to be important and of greatest interest to InfoWorld readers. InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content. Send all inquiries to newtechforum@infoworld.com.