Asset Tracking Performance Follow

The performance of an Aruba Asset Tracking deployment can be defined with the following 3 terms: Accuracy, Latency, and Stability. This document breaks down what you can expect for each.

Accuracy

Definition

The distance between the calculated position of an Asset Tag and its actual real-world position.

Example

If the API produces a position for an Asset Tag that is 5 meters away from where the Asset Tag is physically located, then the "accuracy" of that Tag is 5 meters.

Expected Behavior

The accuracy of an Asset Tracking deployment is directly related to the density of the AP deployment. The more density your AP deployment, the better the accuracy will be. A typical Aruba Wi-Fi deployment is roughly 15 meters, but we strongly recommend a 10 meter deployment for Location Services.

A 10 meter deployment can expect an average of 2 to 5 meters of accuracy with 90% of Tags positioned within 6 meters of their real positions.

A 15 meter deployment can expect an average of 3 to 9 meters of accuracy with 90% of Tags positioned within 12 meters of their real positions.

Test Results (meters)

The graphs below show this same data grouped by accuracy rounded to the nearest meter.

Latency

Definition

The time it takes for the calculated position of an Asset Tag to reflect the physical position of the Asset Tag after the Asset Tag was moved to a new physical position.

Example

If the API indicates a Tag to be at position A, and then that Tag is physically moved to position B, and then it takes the API 3 minutes to indicate that the Tag is at position B, then the "latency" of that Tag is 3 minutes. This does not include the time while the Tag was in motion; the measure of latency begins when the Tag arrives at its new physical position.

Expected Behavior

The average latency is between 2 and 4 minutes with 90% of Tags position updating within 4 minutes of arriving at their new physical position.

Stability

Definition

Also known as "jumpiness", asset tracking stability is the average change in calculated position of a stationary Asset Tag over a period of time.

Example

If an Asset Tag is physically at rest for an entire day, and the calculated position moves an average of 2 meters during the day, then the stability is said to be 2 meters.

Expected Behavior Summary

A 10 meter deployment can expect an average of 0 to 2 meters of instability with 90% of Tags maintaining a position within 3 meters of the average calculated position.

A 15 meter deployment can expect an average of 0 to 1 meters of instability with 90% of Tags maintaining a position within 1 meters of the average calculated position.

Expected Behavior Details

Position Calculations

When an AP detects an Asset Tag, it records the RSSI value, which is a measurement of signal strength, and passes that measurement on to the server. This process is known as an "observation" and the AP is known as an "observer". Each Asset Tag position is calculated using the strongest 3 observations in the past 2 minutes.

For example, if a given Asset Tag is observed by 10 observers in a 2 minute window, then the observations are sorted by RSSI value, and the top 3 strongest signals are used to calculate the position.

Because Meridian uses trilateration, the position of the Asset Tag will always be inside the boundary of the 3 APs. If only 2 APs have observed the Asset Tag, then the calculated position will be somewhere on the line between those 2 APs. If only 1 AP observes the Asset Tag, then the calculated position will be the position of the AP.

Testing Environment

These tests were taken in a standard open office layout with conference rooms in the center of the floor and cubicles along the outer edges. The APs were mounted to the ceiling approximately 3 meters above the floor. Only a single floor in the building was used; multi-floor deployments can introduce other issues not addressed by these tests.

The image below represents the data used for test results in this document. The smaller orange circles are APs and the larger green and orange target circles are Control Tags. The APs with black borders were included for the 10 meter test and excluded for the 15 meter test. You can download the corresponding test data in the attached Excel spreadsheet below.