← back to blog

Nearby Wi-Fi And Bluetooth Scans That Tie Phones Together

Open the Wi-Fi settings on any phone in an apartment block and you’ll see a list of networks, often ten or twenty of them, most belonging to the neighbours. Each one has a name you can read and a hardware address you usually can’t see, called a BSSID, which is unique to that router’s radio. The list your phone sees is a fingerprint of the exact spot where it’s sitting. Move to the next room and the signal strengths change. Move to the next building and most of the names change. Now put two phones side by side on a shelf and they’ll see almost the same list, with almost the same strengths, every time they look. An app that’s allowed to read that list on both phones doesn’t need a shared account, a shared SIM or a shared IP to conclude the two phones are in the same place. This piece covers what apps can actually see about nearby Wi-Fi and Bluetooth, what permissions that takes on Android and iOS, why it matters for anyone running several phones in one location, and what can and can’t be done about it.

Why the radio neighbourhood is a location

Wi-Fi routers and Bluetooth devices broadcast constantly. A router sends out beacons several times a second announcing its network name and its hardware address. Bluetooth devices like earbuds, watches, speakers, cars and other phones advertise their presence so they can be found and connected.

Any phone with its radios on can hear these broadcasts. The combination of which ones it hears, and how strongly, depends on where the phone physically is. Walls, floors and distance all change what reaches it.

This has been used for location for years. The location services on both major mobile platforms combine GPS with Wi-Fi and cell tower information to work out where a phone is, especially indoors where GPS is weak. They do it by comparing the Wi-Fi access points a phone can hear with large databases of where those access points have been seen before. That’s why a phone can show your position inside a shopping centre with no GPS signal at all.

So the radio neighbourhood is more than a vague hint. It’s one of the main ingredients of precise location on a modern phone.

For linking accounts, the database isn’t even needed. If two phones report the same set of nearby access points with similar signal strengths, they’re near each other, whatever those access points are and wherever they are in the world. Two phones that report the same set every day are in the same place every day.

What Android apps can see about Wi-Fi

Android controls access to Wi-Fi information through permissions, and the rules have tightened a lot over the last several versions.

Reading the list of nearby Wi-Fi networks, with each one’s name, hardware address and signal strength, requires location permission on modern Android, and location services must be switched on. Android treats a Wi-Fi scan as equivalent to a location lookup, because that’s effectively what it is.

Android 13 added a separate nearby Wi-Fi devices permission for apps that need to find and connect to nearby devices without using the information for location. Apps that use it are supposed to declare that they won’t derive location from it. That declaration is a rule for developers, and how well it’s followed isn’t something a user can verify from the outside.

Reading details of the network the phone is currently connected to, including its name and hardware address, also requires location permission on current versions. Without it, an app typically gets a placeholder instead of the real network name.

Android also has a setting, usually under location or Wi-Fi settings, called Wi-Fi scanning. When it’s on, the system and apps with location permission can scan for networks even when Wi-Fi itself is switched off. Many people have no idea it’s enabled.

So on Android, an app that has been granted location permission can usually see the full Wi-Fi neighbourhood. An app without it generally can’t, apart from the narrower nearby devices route.

What Android apps can see about Bluetooth

Bluetooth follows a similar pattern.

On Android 12 and later, scanning for nearby Bluetooth devices requires a Bluetooth scan permission. Apps can declare that they don’t use scan results for location, but as with Wi-Fi, that’s a declaration rather than something the user can check. On earlier Android versions, Bluetooth scanning required location permission directly.

A Bluetooth scan returns the devices advertising nearby. Modern phones and many accessories use randomised Bluetooth addresses that change periodically, specifically to make tracking harder. But the advertisements can still carry other information, like device names, manufacturer data or beacon identifiers, depending on the device.

Beacons are a special case. Shops, venues and some public places install small Bluetooth beacons that broadcast a fixed identifier, so apps can tell when a phone is near a particular spot. Those identifiers are designed to be stable and recognisable.

Then there’s the phone’s own name. Every Android phone has a Bluetooth device name, often defaulting to the model, like “Galaxy S21”. Some phones append a number when several share a name. If a phone is discoverable, other phones nearby can see that name.

There’s also a Bluetooth scanning setting, similar to the Wi-Fi one, that lets apps and services with permission scan for Bluetooth devices even when Bluetooth is switched off.

For a phone fleet, Bluetooth is often the forgotten radio. Wi-Fi gets thought about because it carries traffic. Bluetooth sits on by default, doing nothing useful, while advertising and listening.

What iPhones allow

Apple is stricter. Ordinary iOS apps can’t scan the list of nearby Wi-Fi networks at all. That capability is reserved for narrow cases with special approval from Apple.

An iOS app can read the name of the network the phone is currently connected to, but only under specific conditions, typically including location permission and a particular entitlement. For most apps, the connected network name isn’t available.

Bluetooth scanning on iOS requires the user to grant Bluetooth permission, which iOS asks for explicitly. Apps can then scan for nearby Bluetooth Low Energy devices and read their advertisements. Apple also has its own beacon framework for detecting specific registered beacons, tied to location permission.

And location permission on iOS gives apps a position computed from GPS, Wi-Fi and cell data by the operating system. The app doesn’t see the Wi-Fi list, but it sees the result, which is a location precise enough to put two phones in the same building.

So on iPhones the raw Wi-Fi neighbourhood is mostly hidden from apps. Co-location still shows up through location permission and Bluetooth.

Most cloud phones and phone farms run Android, so the Android rules are the ones that usually matter. But the principle carries over. Any permission that reveals location, directly or through radio scans, reveals that phones in one place are in one place.

Cell towers and the other location sources

Wi-Fi and Bluetooth are the sharpest radio signals, but they aren’t the only ones that say where a phone is.

Every phone on a mobile network is attached to a cell, and it can usually see several neighbouring cells as well. On Android, an app with location permission can read the identity of the serving cell and details of nearby ones. Phones in the same room almost always attach to the same cell and see the same neighbours. That’s looser than Wi-Fi, since a cell covers a much larger area and thousands of strangers share it, but it adds to the picture when combined with everything else.

GPS is the obvious one. An app with precise location permission gets coordinates from the operating system, which blends GPS, Wi-Fi and cell data. Indoors, GPS is often weak, so the position leans heavily on Wi-Fi. Phones in one room report nearly the same coordinates, with similar accuracy estimates.

Then there’s IP geolocation, which needs no permission at all. Every app sees the IP address of the connection, and the server can look up roughly where it is. For phones on mobile data, that’s usually the carrier’s city or region rather than a building, so it rarely separates or joins phones on its own. For phones on the same Wi-Fi, it’s the same address, which joins them directly.

Time zone, language and region settings also narrow things down, though not to a room.

So denying Wi-Fi scanning doesn’t make a phone placeless. It removes the most precise radio signal. Location permission and cell data remain, and so does the IP, which is why the permission and radio steps in this piece go together rather than one at a time.

Knowing two phones are in the same place doesn’t prove they belong to the same person. Families share homes. Colleagues share offices. Thousands of strangers share a shopping centre. Platforms know that, and co-location on its own is a weak signal.

It becomes meaningful through repetition and combination.

Repetition is the first. Two phones that happen to be in the same café once are strangers. Two phones that report the same Wi-Fi neighbourhood every day, morning and night, including weekends, live or work together. Over weeks, a platform can distinguish passing overlap from a shared base.

Tightness is the second. Seeing the same shopping centre’s networks is loose overlap. Seeing the same set of networks with nearly identical signal strengths means the phones are within a few metres of each other. Being connected to the same router’s hardware address is even tighter, since it means the phones are on the same network.

Combination is the third. Co-location plus similar account creation dates, plus similar behaviour, plus overlapping contacts or interests, adds up. Each on its own is explainable. Together they describe a cluster.

Some platforms are known to use location signals in features like friend suggestions, which is one reason people sometimes see suggestions for people they’ve only been near. How much weight any platform gives co-location for linking or enforcement isn’t public, and I won’t guess at it.

What’s safe to say is that if accounts are meant to be unrelated, being visibly in the same spot every day is one more thing that makes them look related.

The rack problem

This is where phone fleets run into a limit that can’t be engineered away completely.

A rack of phones in one room is, physically, a rack of phones in one room. They hear the same Wi-Fi networks from the neighbours, at nearly the same strengths. They hear the same Bluetooth devices. If any of them use Wi-Fi, they’re likely on the same router. If any of them use GPS or network location, they report nearly the same coordinates.

Any app on those phones that’s allowed to see location or scan radios will see phones that are together. That’s simply true, and no setting changes the physics.

What can change is how much of it apps are allowed to see, and how much the phones broadcast. Those are the practical levers, covered in the next few sections.

It’s worth being honest that some apps need location to work at all. Delivery apps, ride apps, dating apps and some local marketplace apps require it. For those, co-location of phones in one rack is going to be visible. The choice there is between accepting that, and running the accounts on phones in separate places.

For most other apps, location isn’t required. The app may ask for it, and a real user might grant it, but the account works without it. That’s where denying permissions makes the biggest difference.

What a real household looks like

It helps to know what normal co-location looks like, since that’s the baseline a fleet is compared against.

A typical home has a handful of phones on one Wi-Fi network. Parents, children, perhaps a flatmate or a visiting relative. They’re different models, bought at different times, set up by different people, with different apps. They come and go through the day. Some leave for work or school and appear on other networks. Their owners have different contacts and different interests.

An office has more, maybe dozens, but they arrive in the morning and leave in the evening, and at weekends most of them are elsewhere.

A fleet shelf looks different on almost every axis. The number of phones can be larger than any household. They may be the same model, bought together. They were often set up in the same week, by the same person, with the same apps. They never leave. They’re there at 3am on a Sunday exactly as they are at noon on a Tuesday.

None of those differences requires the app to know anything about you personally. They fall out of comparing the fleet’s radio neighbourhood and presence pattern with what ordinary homes and offices produce.

That’s why the scale question matters. Two or three phones in one place, with different models and setup dates, look like a household even if an app sees them together. A large number of identical, stationary phones that all see the same routers looks like exactly what it is.

If you can’t hide co-location for some apps, keeping the numbers in each location small makes the overlap look much more ordinary.

Your own devices in the room

There’s one more thing the radios pick up that people rarely think about, and it’s the operator’s own devices.

If you work in the same room as your fleet phones, their Bluetooth scans hear your phone, your earbuds, your watch, your laptop, and perhaps your car when it’s parked outside. Many of those advertise names. Some are named after you by default, using your first name followed by the device type, depending on the device and how it was set up.

Your own phone, meanwhile, is signed into your personal accounts. It sits on your personal carrier or your home Wi-Fi. It travels with you.

So an app on the fleet phones that scans Bluetooth, and an app on your personal phone that does the same, can both see each other’s surroundings. The fleet phones hear your devices every day. Your phone is regularly near the fleet phones. That’s a bridge between the fleet and your own identity, formed entirely through radio.

Local networks add another route. Devices on the same Wi-Fi can often discover each other through local network announcements, including smart TVs and printers, speakers too, many with descriptive names. iOS asks users for local network permission for this reason, and apps that get it can see what else is on the network.

The fixes follow from the rest of this piece. Keep fleet phones off your home Wi-Fi. Keep their Bluetooth and scanning off. Give your personal devices neutral names that don’t include your name. And if you can, don’t keep your personal phone in the same room as the fleet all day.

Denying the permissions that matter

The single most effective step is to decide, app by app, which permissions each one actually needs, and deny the rest.

Location is the big one. If an app doesn’t need location to function, deny it. On Android, that also stops it from scanning Wi-Fi networks and, on older versions, Bluetooth devices. On current Android, choose “don’t allow” rather than “only while using the app” where the app still works.

Nearby devices is the next. On Android 12 and later, Bluetooth scanning and some Wi-Fi functions sit behind the nearby devices permission. Most social, messaging and marketplace apps don’t need it. Deny it unless something breaks.

Precise versus approximate location is a middle option on recent Android and iOS. If an app insists on location but only needs a rough area, granting approximate location gives it a general region rather than a precise spot. That still reveals the city, but not that two phones are on the same shelf.

Check the permissions regularly. App updates can add new permission requests, and some apps ask again after a while. On Android, the permission manager in settings shows every app that has each permission, which makes it quick to audit a whole phone.

A real user might grant an app location without thinking. Denying it is also something plenty of real users do, especially privacy minded ones, so it’s an ordinary thing to see from the platform’s side.

Turning off the background scanning

Permissions control apps. The phone’s own scanning settings control whether the radios are listening in the background at all, and they’re easy to overlook.

On Android, look in location settings for the scanning options, often labelled Wi-Fi scanning and Bluetooth scanning. When they’re on, apps and system services with location permission can scan even when Wi-Fi or Bluetooth appears to be off. Turn both off on fleet phones.

Turn off Bluetooth itself unless something on the phone needs it. A phone in a rack almost never needs Bluetooth. Leaving it on means it’s both advertising itself and available to scan.

Turn off Wi-Fi if the phone uses mobile data. A phone on a SIM doesn’t need Wi-Fi for its traffic. With Wi-Fi off, and Wi-Fi scanning off, apps can’t see the Wi-Fi neighbourhood even if they have location permission. Real people often leave Wi-Fi on, so some apps may see this as slightly unusual, but plenty of people turn Wi-Fi off when they’re out.

Make the phone non discoverable over Bluetooth, and consider changing its Bluetooth device name from the default model name. Several phones in one place advertising as the same model with numbers appended is its own small pattern.

Location services themselves can be switched off entirely on phones where no app needs them. That removes GPS, Wi-Fi and cell based location from every app at once.

If phones in a fleet connect to Wi-Fi, the network they connect to matters more than anything they merely hear.

Every phone connected to one router shares that router’s hardware address as its connected network, which apps with location permission can read. They also share the router’s public IP, which every app sees without any permission at all. And they’re on the same local network, which some apps can probe.

That’s a much tighter link than overlapping scan results. It says the phones are on the same home or office network, which is much more specific than being nearby. For accounts meant to be unrelated, it’s close to the strongest non device link there is, alongside sharing an IP through a proxy.

This is one reason phone fleets built for multi account work usually run each phone on its own mobile data connection through its own SIM. Each phone’s traffic leaves through the carrier, on the carrier’s shared addresses, rather than through one router’s single public address. The phones may still be in one room, but they aren’t on one network.

If a phone does need Wi-Fi for some reason, like an initial setup or a large download, connect it to a network that no other fleet phone uses, and disconnect it afterwards. Don’t leave several phones on the same Wi-Fi as their normal state.

Hotspots are a version of the same thing. One phone sharing its connection with others makes them all look like they’re on one network, because they are.

A worked example on one shelf

Here’s a hypothetical to show how the pieces combine.

An operator has six Android phones on a shelf in a spare room. Each has its own SIM and uses mobile data. For convenience, all six are also connected to the home Wi-Fi, so updates download faster. Bluetooth is on, with default device names. Every app was granted every permission it asked for during setup.

A social app installed on all six has location permission. On each phone, it can read the connected Wi-Fi network’s hardware address. All six report the same one. It can scan nearby networks and sees the same fifteen neighbours’ routers at almost identical strengths on every phone. When the phones prefer Wi-Fi over mobile data, all six reach the app from the same home IP. Nearby Bluetooth scans on each phone see the other five phones advertising similar default names.

To the app, those six phones are in one room, on one network, every day.

Now the same shelf, set up differently. Each phone uses only its own mobile data, with Wi-Fi off and Wi-Fi scanning off. Bluetooth and Bluetooth scanning are off. Location permission is denied to every app that works without it, and location services are off on phones where nothing needs them.

The phones are still on one shelf. The app just can’t see it any more. What it sees is six phones on carrier addresses, with no radio neighbourhood to compare.

Placement when location can’t be avoided

For accounts on apps that need location, the permissions approach doesn’t help, and placement becomes the only real lever.

If two accounts have to use an app that requires precise location, and they’re meant to look unrelated, they need to be in different places. Different buildings is the minimum for them to stop looking co-located in any tight sense. Different areas of a city makes the overlap look like ordinary strangers.

That’s a real operational cost, and it’s worth being honest that not every operator can or should take it on. Sometimes the right answer is to run fewer accounts on location dependent apps, or to accept that those accounts are visibly co-located and make sure nothing else links them.

It also matters that the location fits the account. An account on a phone in Singapore, on a Singapore SIM, should be a Singapore account. Spoofing GPS to claim somewhere else, on a phone whose Wi-Fi scans and cell towers say Singapore, creates a mismatch between the claimed position and everything the radios can hear. Location apps compare those, and a GPS position that disagrees with the surrounding networks is one of the classic signs of a spoofed location.

So for location dependent work, the honest options are separate physical placement, fewer accounts, or accepting the link. Faking the location adds a contradiction instead of removing one.

What to check on each phone

Here’s a routine to run on every phone in a fleet. It takes a few minutes per device.

Open the permission manager and look at location. List every app that has it. Revoke it from any app that works without it. Switch remaining apps to approximate location where possible.

Look at nearby devices, sometimes shown as Bluetooth or nearby Wi-Fi permissions. Revoke it from apps that don’t need it.

Open location settings and check the Wi-Fi scanning and Bluetooth scanning options. Turn both off.

Check whether Wi-Fi is on and what it’s connected to. If the phone uses mobile data, turn Wi-Fi off. If it must use Wi-Fi, make sure no other fleet phone uses the same network.

Check Bluetooth. Turn it off if nothing needs it. Change the device name from the default if you prefer, and make sure the phone isn’t discoverable.

Check whether location services need to be on at all.

Write down what you changed, and recheck after any app update or system update, since both can reset permissions or scanning settings.

The honest limit

Radio scans are one of the ways phones reveal where they are, and where they are relative to each other. Controlling permissions and radios reduces what apps can see. It can’t change the fact that phones in one room are in one room, and for apps that require location, that fact stays visible.

I can describe what the permission models on Android and iOS allow, and how co-location becomes a linking signal through repetition and combination. I can’t tell you which apps collect radio scans today, whether they honour their own declarations about not using them for location, or how much weight any platform gives co-location. Those aren’t things an outsider can see.

What I’d take from this is that the physical layer matters. A fleet can be carefully separated in software and still be sitting on one shelf, hearing the same routers, on the same Wi-Fi. For most apps, denying location and switching off the idle radios closes the gap. For the few that need location, where the phones physically are becomes part of the identity.

I run Singapore mobile proxy and cloud phone farms myself, with each phone on its own SIM and mobile data, which is how I think about the shelf problem day to day. If you want the rest of the device and network picture, start on the Multi Account Ops home page.

Get new guides and videos first — join the Telegram channel.

need infra for this today?