Geolocation Sensor

1. Introduction

The GeolocationSensor API extends the Sensor interface [GENERIC-SENSOR] to provide information about the geolocation of the hosting device.

The feature set of the GeolocationSensor is similar to that of the Geolocation API [GEOLOCATION-API], but it is surfaced through a modern API that is consistent across contemporary sensor APIs, improves security and privacy, and is extensible. The API aims to be polyfillable (example) on top of the existing Geolocation API.

2. Examples

Get a new geolocation reading every second:

const geo = new GeolocationSensor({ frequency: 1 });
geo.start();

geo.onreading = () => console.log(`lat: ${geo.latitude}, long: ${geo.longitude}`);

geo.onerror = event => console.error(event.error.name, event.error.message);

Get a one-shot geolocation reading:

GeolocationSensor.read()
  .then(geo => console.log(`lat: ${geo.latitude}, long: ${geo.longitude}`))
  .catch(error => console.error(error.name));

3. Security and Privacy Considerations

Note: Investigate any geolocation-specific security and privacy considerations that are not mitigated by the Generic Sensor API [GENERIC-SENSOR], and define geolocation-specific mitigations.

4. Model

The term geolocation refers to the real-world geographic location of the hosting device represented in geographic coordinates [WGS84].

Note: An implementation can use multiple location information sources to acquire geolocation information, and this specification is agnostic with respect to those sources.

The Geolocation Sensor sensor type’s associated Sensor subclass is the GeolocationSensor class.

The Geolocation Sensor has an associated PermissionName which is "geolocation".

The Geolocation Sensor is a policy-controlled feature identified by the string "geolocation". Its default allowlist is 'self'.

The Geolocation Sensor has an associated virtual sensor type which is "geolocation".

A latest geolocation reading is a latest reading for a Sensor of Geolocation Sensor sensor type. It includes entries whose keys are "latitude", "longitude", "altitude", "accuracy", "altitudeAccuracy", "heading", "speed" and whose values contain the device’s geolocation.

Note: Consider adding a visual of the standard coordinate system for the Earth.

5. API

Issue #62 on GitHub: “Background geofencing”

There are a couple of related issues touching on aspects of supporting background geolocation. Let's use this one to track background geofencing specifically.

For reference, @mkruisselbrink sent a proposed API to the Geolocation WG mailing list back in 2014. That seems like a reasonable place to start if someone wants to pick up this work.

Note that at time of writing no implementers appear to be interested in developing an API for background geolocation or geofencing and so any specification work done in this group will be purely speculative.

5.1. The GeolocationSensor Interface

[SecureContext,
 Exposed=(DedicatedWorker, Window)]
interface GeolocationSensor : Sensor {
  constructor(optional GeolocationSensorOptions options = {});
  static Promise<GeolocationSensorReading> read(optional ReadOptions readOptions = {});
  readonly attribute unrestricted double? latitude;
  readonly attribute unrestricted double? longitude;
  readonly attribute unrestricted double? altitude;
  readonly attribute unrestricted double? accuracy;
  readonly attribute unrestricted double? altitudeAccuracy;
  readonly attribute unrestricted double? heading;
  readonly attribute unrestricted double? speed;
};

dictionary GeolocationSensorOptions : SensorOptions {
  // placeholder for GeolocationSensor-specific options
};

dictionary ReadOptions : GeolocationSensorOptions {
  AbortSignal? signal;
};

dictionary GeolocationSensorReading {
  DOMHighResTimeStamp? timestamp;
  double? latitude;
  double? longitude;
  double? altitude;
  double? accuracy;
  double? altitudeAccuracy;
  double? heading;
  double? speed;
};

Normative changes to the Coordinates interface of the Geolocation API are the following:

Use unrestricted double to not throw TypeError on NaN, +Infinity or −Infinity. (Geolocation API defines NaN as a valid value for heading, yet uses double as its type.)
Use nullable to allow all attributes to default to null. (Geolocation API does not provide synchronous means to access these attributes, thus latitude, longitude, and altitude are not nullable and "must be supported by all implementations".)

To construct a GeolocationSensor object the user agent must invoke the construct a geolocation sensor object abstract operation.

5.1.1. GeolocationSensor.read()

The read() method, when called, must run the following algorithm:

input: options, a ReadOptions object.
output: p, a promise.

Let p be a new promise.
Let options be the first argument.
Let signal be the options’ dictionary member of the same name if present, or null otherwise.
If signal is aborted, then reject p with signal’s abort reason and return p.
Let geo be the result of invoking construct a Geolocation Sensor object with options. If this throws a DOMException, catch it, reject p with that DOMException, and return p.
Invoke geo.start().
If signal is not null, then add the following abort steps to signal:
1. Invoke geo.stop().
2. Reject p with signal’s abort reason and abort these steps.
Run these steps in parallel:
1. If notify error is invoked with geo, invoke geo.stop(), and reject p with the DOMException passed as input to notify error.
2. If notify new reading is invoked with geo, then resolve a geolocation promise with geo and p.
Return p.

Implementations can reuse the same promise for multiple concurrent calls within the same browsing context if the arguments passed to read() are the same.

To resolve a geolocation promise means to run the following steps:

Let reading be a GeolocationSensorReading.
For each key → value of latest geolocation reading:
1. Set reading[key] to value.
Invoke geo.stop().
Resolve p with reading.

5.1.2. GeolocationSensor.latitude

The latitude attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "latitude" as arguments. It represents the latitude coordinate of the geolocation in decimal degrees [WGS84].

5.1.3. GeolocationSensor.longitude

The longitude attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "longitude" as arguments. It represents the longitude coordinate of the geolocation in decimal degrees [WGS84].

5.1.4. GeolocationSensor.altitude

The altitude attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "altitude" as arguments. It represents the geolocation in meters above the WGS 84 ellipsoid [WGS84].

5.1.5. GeolocationSensor.accuracy

The accuracy attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "accuracy" as arguments. It represents the accuracy of the latest reading["latitude"] value and latest reading["longitude"] value in meters with a 95% confidence level.

If the latest reading["latitude"] value is null or latest reading["longitude"] value is null, it must return null.

5.1.6. GeolocationSensor.altitudeAccuracy

The altitudeAccuracy attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "altitudeAccuracy" as arguments. It represents the accuracy of the latest reading["altitude"] value in meters with a 95% confidence level.

If the latest reading["altitude"] value is null, it must return null.

5.1.7. GeolocationSensor.heading

The heading attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "heading" as arguments. It represents the direction of travel in degrees counting clockwise relative to the true north in the range 0 ≤ heading ≤ 360.

5.1.8. GeolocationSensor.speed

The speed attribute of the GeolocationSensor interface returns the result of invoking get value from latest reading with this and "speed" as arguments. It represents the magnitude of the horizontal component of the velocity in meters per second.

6. Abstract Operations

6.1. Construct a geolocation sensor object

input: options, a GeolocationSensorOptions object.
output: A GeolocationSensor object.

Let allowed be the result of invoking check sensor policy-controlled features with GeolocationSensor.
If allowed is false, then:
1. Throw a SecurityError DOMException.
Let geo be the new GeolocationSensor object.
Invoke initialize a sensor object with geo and options.
Return geo.

7. Automation

This section extends Generic Sensor API § 9 Automation by providing Geolocation Sensor-specific virtual sensor metadata.

The per-type virtual sensor metadata map must have the following entry:

key: "geolocation"
value: A virtual sensor metadata whose reading parsing algorithm is the geolocation reading parsing algorithm.

7.1. Geolocation reading parsing algorithm

input: parameters, a JSON Object
output: A sensor reading or undefined

Let reading be a new sensor reading.
Let keys be the list « "latitude", "longitude", "altitude", "accuracy", "altitudeAccuracy", "heading", "speed" ».
For each key of keys
1. Let value be the result of invoking parse single-value number reading with parameters and key.
  1. If value is undefined.
    1. Return undefined.
2. Set reading[key] to value[key].
Return reading.

8. Use Cases

8.1. Categorization of use cases

The mentioned use cases can roughly be grouped into four categories:

Foreground operations:
- Getting a one-off geolocation update.
- Getting continuous geolocation updates (aka. foreground geotracking).
Background operations:
- Getting continuous geolocation updates (aka. background geotracking).
- Getting a one-off geolocation fence alert (aka. background geofencing).

Note: Only the foreground operations were possible with [GEOLOCATION-API], the background operations are completely novel.

Core constraints when obtaining the gelocation are accuracy (how close to the actual position of the user is the determined position) and latency (how long does the user want to wait for a result). Both are tradeoffs: one can trade faster results for lower accuracy and vice versa.

A common theme is to first obtain a rough estimation that then gets refined over time, for example based initially on surrounding WiFi signals (which is fast to obtain) and then eventually based on precise GPS data (which may take some time to find a signal).

In the following, we list use cases based on the previously defined categories.

Note: The categories are not mutually exclusive and overlaps exist. A task might start in the foreground, then continue in the background (for example, while the user quickly responds to an incoming email), and then eventually terminate in the foreground when the user multitasks back.

8.2. Foreground—One-off geolocation update

8.2.1. Locate a user on a map

A mapping application can use the Geolocation Sensor API data of a user to locate them on the map, essentially responding to the question "Where am I right now?"

8.2.2. Find points of interest in the user’s area

Someone visiting a foreign city could access a web application that allows users to search or browse through a database of tourist attractions. Using the Geolocation Sensor API, the web application has access to the user’s approximate current position and is therefore able to rank the search results by proximity to the user’s location.

8.3. Foreground—Continuous geolocation updates

8.3.1. Up-to-date local information

A widget-like web application that shows the weather or news that are relevant to the user’s current area can use the Geolocation Sensor API to register for location updates. If the user’s position changes, the widget can adapt the content accordingly.

8.3.2. Alerts when points of interest are in the user’s vicinity

A tour guide web application can use the Geolocation Sensor API to monitor the user’s position and trigger visual or audio notifications when interesting places are in the vicinity. An online task management system can trigger reminders when the user is in the proximity of landmarks that are associated with certain tasks. This use case assumes active usage of the application in the foreground.

8.3.3. Show the user’s position on a map

A user finds themselves in an unfamiliar area. They want to check their position so they use their handheld device to navigate to a web-based mapping application that can pinpoint their exact location on the map using the Geolocation Sensor API. They then ask the web application to provide driving directions from their current position to their desired destination, essentially responding to the question "Where am I going?".

8.4. Background—Continuous geolocation updates

A social networking application allows its users to automatically tag their status updates with location information. It does this by monitoring the user’s position with the Geolocation Sensor API. Each user can control the granularity of the location information (e.g., city or neighborhood level) that is shared with the other users. Any user can also see their network of friends and get real-time updates about their current location, granted they have opted in to their location data being shared. This use case intentionally conflates foreground location tagging and background location sharing.

A mapping application can help the user navigate along a routeby providing detailed turn-by-turn directions. The application does this by registering with the Geolocation Sensor API to receive repeated location updates of the user’s position. These updates are delivered as soon as the implementing user agent determines that the position of the user has changed, which allows the application to anticipate any changes of direction that the user might need to do. The application can be in the foreground, but likewise can be backgrounded, for example, when the user turns their device off to save battery on a long highway route section without side roads.

8.4.3. Tracking sports activity

A web application can track a user’s sports activity, for example, a marathon run or a bicycle race. Therefore, the application does not need to be on the screen, but would be backgrounded while the user performs their activity, maybe with their handheld device strapped to their arm.

8.4.4. Real estate search

A web application on a handheld device can notify a user of interesting available properties in a neighborhood they are passing by where the property fits the user’s previously specified search criteria, for example, 3 bedroom apartments with balcony. This is based on the assumption that the number of possible matches is high, that is, impossible to realize with geofences as the amount of required geofences would be too high.

8.4.5. Cultural spaces

In cultural spaces like museums, festivals, exhibitions, etc., web applications can react to where the user is or what they are seeing to cue different experiences and content.

8.5. Background—One-off geolocation fence alert

8.5.1. Reminder and to-do applications

Reminder and to-do web applications can use a geofence to remind the user to do something when they cross it, for example, to buy milk when they are near the supermarket.

8.5.2. Travel applications

Travel applications can show venue specific data like WiFi passwords, the user’s booking confirmation etc. only within geofence boundaries.

8.5.3. Ticketing or booking confirmations

Ticketing or booking applications can bring up a ticket notification with a QR or bar code once the user is near the venue of a concert or sports event or when they reach their rental car counter or similar.

Users can be informed if their designated driver reaches a pre-defined pickup point.

8.5.5. Retail special offers

Given their previous consent, a user with a retailer’s web application installed on their handheld device can be alerted about special offers or location-based coupons when they are in vicinity of a physical presence of the retailer. Further, the in-store experience can be enriched, for example, the retailer can let the user know something they have looked at before is actually available for pick up nearby.

9. Acknowledgements

Thanks to Tobie Langel for the work on Generic Sensor API and the early geolocation.js polyfill.

Thanks to Kenneth Rohde Christiansen for the geolocation-sensor.js polyfill and design proposals during the incubation phase.

Special thanks to the Geolocation Working Group participants for their work on the Geolocation API from 2008 until 2017.