IndoorLocation using the Cisco Meraki wifi infrastructure
Clone or download

README.md

Indoor Location with Cisco Meraki

Cisco Meraki is a network, and in particular WiFi, infrastructure which allows to do indoor positioning in multiple ways.

There are 3 possibilities to use Meraki for Indoor Location:

  1. Get device location from the infrastructure: the device's location is computed by the Meraki cloud based on the signal strength of the signals received by multiple access-points.
  2. Have the device computing its position based on the WiFi signal strength of the surrounding access points.
  3. Have the device computing its position based on the bluetooth signal strength of iBeacons embedded in the surrounding access points.

Option 2 requires the mobile phone to be able to scan for WiFi access point, which is not allowed on iOS. Because of this limitation, the solution is not an option for most general public application, but could be used in some enterprise cases when only android phones are used.

Here are some pros and cons between options 1 and 3:

Infrastructure iBeacons
Gets the position of all WiFi enabled devices An app need to be installed on the device
User need to be connected to the WiFi to be able to retrieve its position User does not need to be connected to the WiFi
A new location is computed on average every minute and depends on phone state A new position can be computed every second

Indoor Location from the infrastructure

Every WiFi-enabled device coming in the vicinity of the wifi infrastructure will be positioned by Meraki. The network probe requests coming from the phones will be intercepted by the different access points and their signal strength will serve to triangulate the position. Each device is identified by its MAC address.

To protect the privacy of the user, most modern phones running iOS and Android will use random MAC addresses in their probe requests. Therefore, one need to pay attention while analyzing the received data that the MAC address of a single device will change over time.

Probe requests are sent by the phone to discover the network around them. Depending of its state (active, sleeping, energy-saving, in communication, ...), the phone will not necessarily always send a probe request at the same interval. We can expect on average to have a probe request every minute but the infrastructure has no control on it.

On iOS and Android, it is impossible for an app to retrieve the MAC address of the device, also for privacy reasons. So our only option to identify the location of a specific device is using the local IP address of the device.

The procedure works as such:

  • The location of the device is computed by the Meraki cloud
  • Those locations are sent as notifications to the IndoorLocation server (see below)
  • The server decodes the location and stores it in a cache (Redis or in-memory) using the local IP or MAC as key. It can also send it to any other system or database for further processing.
  • The device connects, thanks to web sockets, to the IndoorLocation server and provides its local IP as userId.
  • The server listens for changes in the cache and sends the right location to the right user matching the local IP

Floor and aligment

The Meraki cloud provides a device location as a latitude / longitude pair. However, there are 2 reasons why an extra processing is required:

  • The provided location is missing a floor, which is critical for Indoor applications.
  • Most of the time, the alignment of the building floor plans on the world map done in Meraki is approximative and not done with indoor positioning in mind, and therefore will not match with the indoor map in your application.

A correction phase is then required and works like this:

  • Using the configurator provided below, the Meraki configuration (the floor plans and their alignments) are injected in Mapwize as layers.
  • In Mapwize, we configure a floor for each imported floor plan.
  • If required, we move the floor plan in Mapwize according to our indoor map.
  • Using the configurator provided below, we get a configuration file that will be used later by the listener to process every location received from Meraki.

App SDK

The configurator and the server provided in this repository are running on the server side.

If you want to get the user's position in your app, and for example display it on the map, you will need to use the SocketIndoorLocationProvider modules which are available for both iOS and Android

Indoor Location using iBeacons

All Meraki access points can be configured to emit a iBeacon bluetooth signal. Those signals can be heard by the mobile phones and used to compute the device's location. See the Bluetooth Beaconing section in the Meraki documentation

In order to do so, an iBeacon SDK needs to be included in your app. IndoorLocation proposes a BasicBeaconIndoorLocationProvider available for both iOS and Android. Other providers exists with more comprehensive features.

In order to use the BasicBeaconIndoorLocationProvider, the steps are the following:

  • Using the configurator (see below), import the floor plan configuration from Meraki to Mapwize. Set the floor of each floor plan and re-align it if necessary.
  • Import the position of the access points in Mapwize with their iBeacon UUID, major and minor settings.
  • Add the BasicBeaconIndoorLocationProvider in your mobile app and make sure the API key used has read access to your venue.

Configurator

A configurator tool is provided to import data from Meraki to Mapwize and create the configuration file required to run the Meraki listener server.

The configurator is split in 3 steps described below.

Getting data from Meraki

Unfortunately, at this point, Meraki is missing the required API to extract any floor plan information. Therefore, we'll need to find ways around that. If you have the chance to speak to someone from Meraki, please don't hesitate to up-vote the idea of a floor plan API :-)

Here is the description of how to get the required data from Meraki.

Getting the floor plans

  • Log in to your Meraki dashboard
  • Open the Developer Console to log all requests
  • In the Wireless menu, navigate to Map & floor plan
  • Go the the Network tab in the developer console, find the floorplans API request (/floorplans) and save the JSON into a file

Getting the list of access points

  • Log in to your Meraki dashboard
  • In the Wireless menu, navigate to Access points
  • On the top right, use the Download As button to download the list of AP as XML

You might wonder why we use XML in this case and not JSON as everywhere else? Simply because in the JSON export, the nodeId of the access points are not exported, while in the floor plans we got above, only the nodeId are available. Meraki God, if you hear us...

Install

The configurator is a command line tool developed in NodeJS. Before running it, you need to have node installed on your machine and install the dependencies using npm install.

Upload Meraki floor plans into Mapwize

This step creates a layer in Mapwize for each floor plan in Meraki. The layer's name is Meraki - followed by the floor plan name.

At first, these layers do not have a floor configured, as the floor data is not available in Meraki. You need to manually edit each layer for updating their floor properties.

As some alignment problems can be observed, you might also have to check that each layer image perfectly overlays the Mapwize layers for a better accuracy.

The command is described below:

./meraki-configurator/merakiFloorplansToMapwize.js --merakiFloorPlansConfig [FILEPATH_FOR_FLOORPLANS_JSON] --mapwizeUser [YOUR_MAPWIZE_EMAIL] --mapwizePwd [YOUR_MAPWIZE_PWD] --mapwizeApiKey [YOUR_MAPWIZE_API_KEY] --mapwizeOrganizationId [YOUR_MAPWIZE_ORGANIZATIONID] --mapwizeVenueId [YOUR_MAPWIZE_VENUEID]

Generate the configuration file

Once the layers are correctly configured in Mapwize, with right floor and alignment, use the command below to generate the JSON configuration file that will be used later by the server.

./meraki-configurator/configureFromMapwize.js --merakiFloorPlansConfig [FILEPATH_FOR_FLOORPLANS_JSON] --mapwizeUser [YOUR_MAPWIZE_EMAIL] --mapwizePwd [YOUR_MAPWIZE_PWD] --mapwizeApiKey [YOUR_MAPWIZE_API_KEY] --mapwizeOrganizationId [YOUR_MAPWIZE_ORGANIZATIONID] --mapwizeVenueId [YOUR_MAPWIZE_VENUEID] --output [OUTPUT_PATH_FOR_LISTENER_CONFIGURATION]

Import the access points to Mapwize

This steps creates beacons in Mapwize based on the Meraki configuration. Please note that the beacons will be added in Mapwize based on the floor plan configuration done in Mapwize, with the floor and alignment given to the floor plans.

./meraki-configurator/merakiAccessPointsToMapwize.js --merakiFloorPlansConfig [FILEPATH_FOR_FLOORPLANS_JSON] --merakiAccessPointsConfig [FILEPATH_FOR_ACCESSPOINTS_XML] --mapwizeUser [YOUR_MAPWIZE_EMAIL] --mapwizePwd [YOUR_MAPWIZE_PWD] --mapwizeApiKey [YOUR_MAPWIZE_API_KEY] --mapwizeOrganizationId [YOUR_MAPWIZE_ORGANIZATIONID] --mapwizeVenueId [YOUR_MAPWIZE_VENUEID]

Server

This is the NodeJS server receiving the Meraki Location notifications and sending them to the app using web socket.

Each notification will be received and processed into an IndoorLocation object. The computed IndoorLocation will be saved in a cache (Redis or in-memory).

Use

We first need to correctly set the configuration parameters directly in the config/all.js file or via environment variables

  • PORT: port used by the server (default: 3004)
  • SECRET: secret defined in the Meraki dashboard to authenticate the POST query (required)
  • VALIDATOR: token defined by Meraki to identify the source (required)
  • MAX_BODY_SIZE: Controls the maximum request body size (default: '50mb')
  • FLOOR_PLANS: serialized JSON with the output of the configurator (required)
  • MAC_ADDRESS_ENABLED: use the MAC address in addition to the IP address as a key in the cache (default: true)
  • REDIS_ENABLED: using Redis if true (default: false)
  • REDIS_HOST: redis host (default: localhost)
  • REDIS_PORT: redis port (default: 6379)
  • REDIS_AUTH: redis password (required if set)
  • MERAKI_NOTIF_TTL: how long a location is kept if no new location is sent for that device in seconds (default: 3600)
  • DOCUMENT_DB_ENABLED: allow to log the indoorLocation along to the Meraki observation into a DocumentDB collection
  • DOCUMENT_DB_ENDPOINT: string connexion to the DocumentDB instance
  • DOCUMENT_DB_PRIMARY_KEY: primary access key to the DocumentDB instance
  • DOCUMENT_DB_DATABASE: DocumentDB database name
  • DOCUMENT_DB_COLLECTION: DocumentDB collection name

Start the server using

npm run start-listener

Notes

To correctly serialize the Meraki configuration, one can execute the command described below:

export FLOOR_PLANS=$(node -e 'var json = require("FILEPATH"); console.log(JSON.stringify(json));')

If you want to put this variable into your clipboard instead, please execute the command below:

node -e 'var json = require("FILEPATH"); console.log(JSON.stringify(json));' | pbcopy

Areas

If a list of areas is provided, the server can tag each device with the area it is in. A device is expected to be in only 1 area, and therefore areas should not overlap.

The list of areas can be provided using the AREA environment variable as a GeoJSON FeatureCollection. Each feature needs to have a geometry of type polygon. Also, each feature should specify in it's properties its name (as string) and floor (as number).

Such list of areas can easily be created using Mapwize Studio. Create places for your areas in your venue and then use the "Export places as geojson" function to download the GeoJson file.

Production deployment with Redis

The in-memory database is great for development and testing but is not suitable for production deployments because it does not persist data in case of server restart and it does not allow to scale the number of instances.

To overcome those issues, we can use a Redis cache.

For the system to work, Redis notifications have to be enabled! It can be done using the following command:

redis-cli config set notify-keyspace-events K$

On some cloud platforms, like Microsoft Azure, the notifications can be enabled from the "Advanced settings" menu.

When Redis is in place, it becomes possible to specialize some instances of the server to be only listening to location notifications. This is particularly useful when the load is mainly coming from the number of devices tracked.

On the other hand, some instances of the server can be dedicated to the socket communication with the devices. This is particularly useful when the load comes form the the number of devices requesting their location.

Storing data

The data can be stored into an external databases for further analysis. MySQL and Azure documentDB are supported at this point.

Please note that the volume of data can be quite large. Dimention your database and your retention period accordingly.

Storing in MySQL

To store the data in MySQL, create a table with the following fields:

CREATE TABLE location (clientMac VARCHAR(20), type VARCHAR(20), latitude DOUBLE, longitude DOUBLE, floor DECIMAL(5,2), accuracy DOUBLE, timestamp INTEGER, apMac VARCHAR(20), rssi INTEGER, ipv4 VARCHAR(20), area VARCHAR(50))

Then use the configuration variables to enable the connection:

  • MYSQL_ENABLED
  • MYSQL_HOST
  • MYSQL_PORT
  • MYSQL_USER
  • MYSQL_PASSWORD
  • MYSQL_DATABASE
  • MYSQL_TABLE

Assets

The assets feature lets you query a list of pre-defined assets with their location using the /assets route.

Define an array of objects in the ASSETS environment variable. The entire array will be returned when doing a GET on /assets. Each asset should be defined as an object that contains its MAC address in the id property.

Packaging

A Docker image has been made available to ease the deployment. The image can be locally built as follows:

docker build -t meraki-indoor-location:latest -f docker/Dockerfile .

Once built, you can start the server with: docker run --rm -it -p 3004:3004 meraki-indoor-location:latest start. All environment variables can be passed to the container via the --env parameter during the container creation.

Contribute

Contributions are welcome. We will be happy to review your PR.

Support

For any support with this provider, please do not hesitate to contact support@mapwize.io

License

MIT