The power of rooms

The Kontakt.io Location Engine Reference Architecture is an AI-based framework designed to achieve accurate, room-level location information for both people and things in indoor spaces. In this article, you'll learn about the system architecture, the power of rooms and room sensors, as well as the more common deployment topology options.

The system architecture

Click on an image to expand

System architecture illustration

Tags

A Tag is a device that transmits telemetry payload over Bluetooth® Low Energy (BLE) technology and is commonly attached to assets or worn by people.

Tags, such as Kontakt.io’s Asset Tag 2 or Smart Badge, can also be equipped with an Infrared (IR) receiver that can receive signals containing room IDs.

Infrastructure devices

An Infrastructure device is a stationary device that is placed onto a ceiling or wall.

These devices include:

BLE-enabled Wi-Fi Access Points. Act as Gateways*; Cisco Catalyst 91xx series.
BLE to Wi-Fi Gateways*. Act as an overlay network when local Wi-Fi Access Points do not have Bluetooth Low Energy (BLE) connectivity; Kontakt.io Portal Light.
Room Sensors. Kontakt.io’s Portal Beam and the Beam Mini act as both a BLE reference and infrared (IR) beacon to improve the location engine accuracy. IR, unlike BLE, does not penetrate the walls allowing for room-level granular accuracy.

* Gateways send data from Tags and Room Sensors to Kio Cloud over their connection to the local Wi-Fi network.

The Kio Cloud

This is Kontakt.io’s cloud-based platform that collects, computes, manages, and delivers location based data services and solutions through its room-based location typology and Location Engine. The Location Engine processes the location typology, Tag and Infrastructure data to calculate and output the location of people and things in an indoor environment.

At Kontakt.io, we are primarily interested in grid-designed buildings. Hospitals and office buildings are good examples, whereas airport hangars and large warehouses are not. The Kio Cloud location topology and Location Engine are built on room logic. Room logic divides any building and floor into rooms and assigns an unique identifier to each room

Example: floorplan as room-level matrix

We believe that within a hospital, office building, and any other grid-designed buildings, both humans and software of any kind best consume location information through the concept of rooms. Room level information allows us to measure how much time nurses have spent with patients, automate workflows from staff safety, nurse-call, hand hygiene or PAR-level inventory management. To support these use-cases, software applications need to know with certainty if someone or something has been inside or outside of the room in real time and over time. More granular data is never needed. Zone based information or a blue dot jumping in and out of a room is not sufficient to support use cases that require room-level certainty.

Rather than a two-dimensional (2-D) space like a map, we consider indoor spaces as a matrix both geometrically and mathematically. From a human perspective, the smallest location unit (a “unit of space”) we consider is a room or in open spaces where a “room” could be. More technically, each such room is a matrix cell. In traditional approaches, location engines calculate a 2-D relative location output which is then benchmarked by a policy engine against geofences to finally deliver a “room” or “zone” localization output for data consumers. The matrix approach instead, reduces computational overhead, humanizes the location output by design and helps provide deeper insight into spaces and interactions between spaces, people, and things. This underlying logic allows the Location Engine to provide room-level accuracy.

The power of Room Sensors

Deploying Room Sensors, such as Kontakt.io's Portal Beam or Beam Mini, within each room significantly improves the room-level accuracy in comparison to any standalone Gateway deployment for both IR-enabled and BLE-only Tags.

For the geeks out there, here are the technical reasons. Kontakt.io’s Machine Learning (ML) powered Location Engine is…

Constantly predicting the whereabouts of Tags against the room-level matrix. This makes the location outputs less error prone to room jumps in a 1:1 comparison to a 2-D location engine.
Using Beams as reference beacons and applying particle filtering ML algorithms as well as unsupervised learning using real-time Infrared (IR) referencing.
Considering the BLE RSSI device specific propagation models that we use to train the engine on specific device antennas. Each device has different characteristics.

In addition, our Location Engine has preset use-cases and device-dynamic location engine window parameters that define how many signals we use to process and predict a location update. This allows us to optimize and balance latency and room-level certainty. The longer the window, the better the room-level certainty and the lower the latency. This is what makes Kontakt.io’s location outputs the most reliable and robust in the industry, no matter the infrastructure.

Deployment topology 101

The deployment topology, from the type and placement of infrastructure devices to the technology of tags, impacts the room level certainty computed by our Location Engine.

The following highlights the four most common deployment topologies. This information is intended to serve as reference when designing a Kontakt.io location-based solution. Each topology is based on the Kio Cloud Location Engine unless otherwise specified.

1. Room Sensors and Cisco Wi-Fi BLE Access Points

This deployment topology delivers the highest level of room-level accuracy for both Tags with BLE-only and those with BLE + Infrared (IR) capabilities.

One Portal Beam or Beam Mini (room sensor) is mounted in each room and throughout the corridors. Beams do not require cabling or Power over Ethernet (PoE).
With the right density of Access Points, additional Portal Lights are not required.

2. Room Sensors and Portal Lights in every room

This deployment topology delivers the highest level of room-level accuracy for both Tags with BLE-only and those with BLE + Infrared (IR) capabilities.

One Portal Beam or Beam Mini (room sensor) is mounted in each room and throughout the corridors. Beams do not require cabling or Power over Ethernet (PoE).
In the absence of Access Points, one Portal Light is installed in each room. To achieve the desired accuracy results, use of power outlets at the same height and position in each room.

3. Portal Lights or Cisco Wi-Fi BLE Access Points in every room (no Room Sensors)

This deployment topology delivers a lower level of room-level accuracy for any type of Tag.

Portal Lights in each room; power outlets at the same height and position in each room should be used.
Best used to establish a first-level of area or zone based location awareness.

Tags	Room level certainty	Use cases
Tags with BLE only; Kio or third-party Tags	90% +	Tracking visibility with 1 minute to 5 minutes latency.

Example: Portal Lights in each room (no Room Sensors)

4. Only Cisco Wi-Fi BLE Access Points and no Kio Cloud Location Engine

This deployment topology delivers the location engine output as defined by Cisco Spaces.

In summary

Kontakt.io's Location Engine Reference Architecture provides reliable location outputs through a three-component system architecture: Infrastructure, Tags, and the Kio Cloud. The infrastructure consists of BLE-enabled Cisco Wi-Fi Access Points, Kontakt.io Portal Lights (BLE to Wi-Fi Gateways), and Room Sensors.

The location engine is designed for grid-based buildings, where rooms serve as the primary location unit, ensuring efficient and accurate location data for various use cases.

Kontakt.io's Machine Learning (ML) powered Location Engine constantly predicts Tag whereabouts and utilizes Beams as reference beacons for improved room-level accuracy. This approach, along with the Kio Cloud's preset use-cases and device-dynamic Location Engine window parameters, makes Kontakt.io's location outputs the most reliable and robust in the industry.