Accurate Navigation Coming to a Phone Near You
At last week’s Google I/O developer festival, engineers from Google’s Android location team previewed some exciting advancements coming soon to Android mobile devices. Two technologies will significantly improve location accuracies:
- Wi-Fi round-trip time (RTT) will enhance indoor location accuracy and enable indoor navigation.
- Global Navigation Satellite System (GNSS) will achieve sub-meter location accuracy outdoors in open-sky scenarios.
Round-trip time (RTT) is a new addition to the Wi-Fi standard (IEEE 802.11mc) that defines the protocol for measuring the time it takes a message to travel between a device and a specified Wi-Fi access point. Alone, RTT may not seem very useful, but given that Wi-Fi signals travel at the speed of light, the time can be converted to distance to calculate the device’s distance from the access point.
Now we’re on to something! By calculating the distance from a device to at least four access points, we can triangulate that device’s position in 3D space.
This triangulation capability brings accurate positioning indoors where GPS doesn’t function. Google’s fused location provider API uses crowd-sourced locations of known Wi-Fi access points to triangulate indoor positions of Android devices.
Currently, this API compares the strengths of signals received from visible Wi-Fi access points to approximate a position—with accuracy to around 10 meters. The new RTT triangulation can achieve location accuracy down to 1 meter.
This is a game-changer for indoor mapping. Navigation apps could guide users down specific hallways and differentiate between building floors and rooms. Shopping apps could point to specific store aisles, and even specific aisle sections, to locate items. The possibilities are virtually endless.
The beauty of RTT is that it may not require much new hardware. Most Wi-Fi access points and radios manufactured within the last three years already have the necessary components, and RTT can be enabled via firmware updates.
Google will be updating its own Wifi routers with RTT-capable firmware later this year and will begin shipping RTT-default-enabled devices soon after. Also, because RTT is not a proprietary Google technology (it’s part of the IEEE Wi-Fi standard), we likely will soon see 1-meter indoor accuracies available on iOS and other devices.
Currently, GPS chips in mobile devices are limited to accuracies of around 5 meters, due to the wavelength of satellite signals. Surveying hardware improves this accuracy by adding Global Navigation Satellite System (GNSS) signals from reference station networks. The different frequencies offered by GNSS reference networks enable receivers to interpolate between GPS signals and enhance measurement precision.
Today’s mobile GPS chips operate on the L1 frequency alone, which limits their precision. A new generation of GPS chips from manufacturers like Broadcom can operate on both L1 and L5 frequencies. Devices built with these new chips can use GNSS networks to enhance their location accuracy.
Google is adding experimental GNSS support to the next version of Android, Android P, due out this fall. New APIs will enable developers to check a device’s GNSS capability and obtain raw readings from the GNSS receiver. Accurate GNSS readings require receivers to be powered on, so Android P includes a new developer option to disable duty cycling (periodic power-downs to save battery) and maximize GPS receiver accuracy.
These advancements point to near-survey-grade location accuracy on Android devices. Earlier this year, Trimble analyzed the performance of such devices with GNSS capability and found it was possible to achieve 2-centimeter accuracy. With small antennas, mobile chips take longer to converge than professional surveying equipment. This enhanced location accuracy on mobile devices will be incredibly useful; possible applications include using GNSS to enable lane-level navigation and traffic analysis.
We’re at least one year away from seeing the widespread adoption of this new generation of mobile GPS chip, and I’m excited to try it out. I can see infinite utility in GIS, especially for asset management and field data collection. I’m also curious what may be achieved by combining GNSS accuracy with the visual positioning capabilities of ARCore and ARKit. We may be looking at a new class of apps able to locate devices with unprecedented accuracy.