How Bluetooth Low Energy took over Bluetooth

By Shailendra Kumar Ram · 2026-09-07 · 12 min read

There is a radio in your earbuds, your car key, your watch, your thermostat and, if you have ever lost your keys, in a small white disc in your bag. All of it is called Bluetooth. Almost none of it is the Bluetooth that was invented in 1994. The protocol doing that work is a different radio, designed by a different company, for a different problem, that ended up wearing the older technology's name.

That substitution is the most interesting thing about Bluetooth Low Energy, or BLE, and it is the part every history of it leaves out. The usual telling is a table of version numbers: 1.0, then 2.0, then 4.0 brought low energy, then 5.0 doubled the speed. It answers what and when, and it never once answers why. So here is why, which turns out to involve a coin cell, a Finnish handset maker losing an argument on purpose, and a tenth-century Danish king who got the job because nobody had time to finish a trademark search.

Why it is called Bluetooth: a king, a book, and a trademark search

Start with the name, because the name explains the ending. In about 1997, after a Home-RF meeting in Toronto, Intel's Jim Kardach spent an evening in the city's pubs with Sven Mattisson, one of the two Ericsson engineers who had built the radio. Mattisson, being Swedish and several drinks in, told him the history of Scandinavia by way of a historical novel he liked, Frans G. Bengtsson's The Long Ships.

Kardach went home interested enough to keep reading. A book on Scandinavian history he had ordered, Gwyn Jones's The Vikings, was waiting for him, and in it he found a picture of a runestone with a figure carved on it, raised by a king called Harald Bluetooth. Harald took the Danish throne around 958 and later brought Norway under his rule, and a technology meant to unite the computing and mobile industries needed a codename for its working group. That is why the radio in your earbuds is named after a tenth-century Danish king.

It was only ever meant to be a placeholder. Two real names were in play. Intel's marketing put forward RadioWire, championed by Simon Ellis. Ericsson and others backed PAN, for Personal Area Network. PAN won the vote four to one.

Then it died. Three weeks later, Ericsson's legal team came back from the trademark search with roughly 1,700 hits on PAN and a risk assessment of, in Kardach's telling, extremely high. RadioWire's own search could not be finished in time for the launch. That left exactly one name that anybody had already cleared, and it was the Viking king. The logo is a bind rune, the Younger Futhark characters Hagall and Bjarkan set on top of each other: Harald's initials.

The most successful wireless brand in history is a placeholder that survived because the alternatives ran out of time.

Two details make it better. A later focus group found that a name nobody remembers, CoMeGo, tested more strongly, but by then Bluetooth had too much industry recognition to be worth changing. And Kardach, who had been promised five hundred dollars for coming up with it, was paid two hundred. His own account of all this repays reading in full; it is the primary source that every retelling quietly paraphrases, usually while getting the book wrong.

What the first Bluetooth was built to do

The radio itself is older than the name. Ericsson dates the idea to 1989, crediting Nils Rydbeck, then chief technology officer of Ericsson Mobile, and the physician and inventor Johan Ullman. Tord Wingren was given the job of specifying it and assembling a team, and Jaap Haartsen and Sven Mattisson were commissioned to design it. Haartsen's work on the short-range radio took shape around 1994.

The brief was narrow and worth stating plainly, because everything that follows is a consequence of it: replace the RS-232 cable, and connect a computer to a wireless headset. That is a streaming voice link between two devices, one of which has a battery its owner charges every day.

Design for that and you make a specific set of choices. You build a connection-oriented protocol, because a headset holds a link open for the length of a call. You hop across 79 channels to survive a crowded 2.4 GHz band. You accept that finding another device is expensive, because a person only pairs a headset once and then forgets about it. Every one of those decisions is correct. The Bluetooth Special Interest Group formed around them in May 1998, with Ericsson, IBM, Intel, Nokia and Toshiba as founders, version 1.0 followed in 1999, and in 2001 the Ericsson T39 became the first widely available phone with a Bluetooth radio in it.

The myth worth correcting

No history of Bluetooth escapes the claim that it rests on Hedy Lamarr's frequency hopping patent. The patent is real, filed in 1941 with George Antheil and granted in 1942, and its mechanism, synchronising transmitter and receiver using something very like a player piano roll, was genuinely inventive.

The causal story is not. Ellison Purington had filed a proposal for wobbling a carrier's frequency to reduce interference in September 1940, a year earlier. And as American Scientist has documented, the seventh claim of the Lamarr and Antheil application, the one that reads most like a general definition of frequency hopping, was rejected by the patent examiner on prior art. The claim that would have made them its inventors is precisely the claim the Patent Office refused.

There is also no evidence that any Ericsson engineer consulted it. Haartsen chose frequency hopping in the 1990s because it was the standard, obvious answer to coexistence in a crowded 2.4 GHz band, which is a duller explanation and the correct one. The myth survives because it is a better story than engineers picking the conventional solution. The companion claim, that Lamarr invented Wi-Fi, is worse: the Wi-Fi anyone uses today does not frequency hop.

The 22.5 milliseconds that made everything else impossible

And then somebody asked whether the same radio could run a door sensor for a year on a CR2032.

The intuitive answer to that question is to turn the transmit power down, send less often, and sleep in between. It does not work, and the reason it does not work is the single most important fact in this entire story.

The power in a classic Bluetooth link is not mostly spent transmitting. It is spent finding out whether anyone is there. Discovery in classic Bluetooth sweeps 32 of its channels, and a scan takes on the order of 22.5 milliseconds. A device that must remain findable does that, repeatedly, for as long as it is deployed. There is no duty cycle low enough to make that free, because the cost is architectural rather than a setting.

Three advertising channels instead of thirty-two, and a scan roughly twenty times shorter. That is the invention. Everything people think of as characteristic of BLE follows from it: the advertising model, the connectionless discovery, the fact that a peripheral's advertising interval is the one knob that decides its battery life, and the reason a scan can come back empty when that interval is set too long. The three advertising channels are even positioned in the gaps between the commonly used Wi-Fi channels, so the most frequently used part of the protocol is the part least likely to collide with the noisiest thing in the band.

An independent peer-reviewed evaluation confirms the structure: 40 channels, three of them for advertising, advertising intervals from 20 ms to 10.24 seconds, plus a random delay of up to 10 ms on each one so that two devices that start together do not stay in lockstep and collide forever.

Nokia builds a different radio

Nokia Research had been working the problem for a while, and in October 2006 it announced the result publicly under the name Wibree. The target was explicit: a device running from a coin cell with an average operating life of about a year, using roughly a tenth of the energy of Bluetooth for the same job.

The launch partners are a good snapshot of who cared. CSR, Broadcom, Nordic Semiconductor, Epson, Suunto and Taiyo Yuden. Nordic is the name to notice, because if you have shipped BLE firmware in the last decade there is a fair chance you did it on their silicon, and they were there before the protocol had its current name.

Nokia also made a decision that decided the outcome. Wibree was designed to run dual-mode, sharing a chip alongside classic Bluetooth, so that a handset maker could adopt it without adding a second radio and a second bill of materials. It was built from the start to coexist with the thing it was going to replace.

Why Wibree had to stop being Wibree

On 12 June 2007, less than a year after announcing it, Nokia handed Wibree to the Bluetooth SIG, where it was renamed Ultra Low Power Bluetooth. A company gave away a working technology it had funded, to a standards body it did not control, and the reason was stated at the time by Eric Janson, then senior vice president for global sales at CSR, in EE Times:

Clearly, Wibree as a standalone specification had a drawback with many potential users seeing it as a Nokia specification.

That is the whole argument in one sentence. A standard owned by one handset maker is not a standard, it is a supplier's product with a specification document attached. Nokia could have kept control of Wibree and watched it lose to whatever the SIG eventually shipped, or it could trade control for universality. It traded, and BLE now ships in billions of devices a year rather than in Nokia phones.

I went looking for the fight in this story and could not find one. The contemporaneous coverage in EE Times, The Register and MIT Technology Review all describe a negotiated, mutually advantageous merge. There are hints that the SIG had low-power work of its own underway, which implies somebody's project was quietly shelved, but nobody is on record objecting. For a standards body absorbing a rival specification, the absence of a public argument is itself unusual.

Bluetooth 4.0 was not version four of anything

The Bluetooth Core Specification 4.0 was adopted in December 2009, with low energy as its headline. The version number implies continuity, and there is almost none. BLE and classic Bluetooth are not interoperable. Different channel plan, different discovery model, different protocol. A 2025 review in Sensors puts it plainly: BLE is independent of BR/EDR and has no compatibility with it, though the two can coexist on the same chip.

So a device advertising "Bluetooth 4.0" was not running an improved version of the 1994 radio. It was running two unrelated radios in one package, one of which had been designed by Nokia three years earlier under a different name and for a different purpose.

Which means the same thing happened twice. In 1998 a Viking king beat RadioWire because his name was the only one that had cleared legal. In 2009 Nokia's radio beat its own brand because Bluetooth's name was worth more than Wibree's technology. Both times, the engineering was settled and the naming decided what happened next.

Apple shipped first, and Android took twenty-one months

A specification is not a platform. BLE existed on paper from the end of 2009 and could not reach a normal person until a phone carried it.

Apple went first. iOS 5 arrived on 12 October 2011 with the Core Bluetooth framework, and the iPhone 4S went on sale two days later as the first widely available phone with a Bluetooth 4.0 radio. The SIG noticed: on 24 October, ten days after the phone shipped, it rebranded low energy as "Bluetooth Smart", a name it later abandoned.

Android's support landed with 4.3 in July 2013, at API level 18. That is the twenty-one month gap everybody quotes, and the real asymmetry is worse than the dates suggest, because 4.3 supported the central role only. An Android phone could scan for peripherals and connect to them. It could not be one. Android has kept rewriting the terms since, most recently when Android 12 split the Bluetooth permissions.

That did not change until Android 5.0, in Google's own words: "Android 4.3 introduced platform support for Bluetooth Low Energy in the central role. In Android 5.0, an Android device can now act as a Bluetooth LE peripheral device." Apple introduced iBeacon in June 2013. An Android phone could not act as a beacon until Android 5.0 arrived in November 2014, so the category Apple created ran on one platform for its first seventeen months.

I have never found a documented explanation from Google for the delay, and I would not guess at one. What I can say from my own work is that the asymmetry outlived the versions: hardware companion apps still tend to be specified against what iOS does and then adapted, and a peripheral that Android sees and iOS does not is still a reliable surprise late in a project.

The uses nobody specified

Here is the pattern in what BLE ended up doing: the applications that mattered were almost never the ones anybody wrote the specification for. The uses that were planned for tended to disappoint, and the ones nobody anticipated went everywhere. Apple introduced iBeacon at WWDC in June 2013, and for two years proximity beacons were going to change retail. Google answered with Eddystone in July 2015. Neither ended where the coverage suggests. Google discontinued Nearby Notifications in December 2018 and deprecated its Beacon Platform two years later, but the Eddystone format itself was never withdrawn and still works. "Google killed Eddystone" is one of those things everyone says that is not quite true.

The fitness wave is remembered slightly wrong too. The original Nike FuelBand, which is the device most people picture when they think of the moment wearables arrived, shipped in February 2012 without Bluetooth 4.0. The FuelBand SE, in November 2013, was the one with BLE in it. The category is older than the radio it is credited to, and the economics of putting one on the market have not got simpler since.

Then in April 2020 the protocol got a job nobody had designed it for. Apple and Google's Exposure Notification system used BLE advertising to have phones record proximity to each other, with rotating identifiers that changed every 10 to 20 minutes so the trail could not be followed. Nobody in 2006 was designing an advertising channel for an epidemic. Whatever you think of how it went, it was the largest deliberate deployment of BLE advertising ever attempted, and it ran on the same three channels Nokia had specified fourteen years earlier.

A year later, Apple's AirTag announcement turned the installed base itself into infrastructure. It is worth being precise about how, because the two halves get conflated: BLE does the finding, and ultra-wideband does the pointing. Any passing iPhone hears the tag's advertisement and reports its location; the U1 chip only matters in the last few metres. BLE solved the scale problem and never solved the precision one.

The spec stopped being an event

Bluetooth 5 was announced in June 2016 and adopted that December, with three headline claims: twice the speed, four times the range, eight times the broadcast capacity. They are not simultaneous, and marketing rarely mentions it. The range comes from trading bandwidth away, so 2x speed and 4x range are alternative operating modes you choose between, not a combined upgrade you receive.

Two of those deserve more attention than they get. Channel Sounding measures distance using phase-based ranging and round-trip timing instead of inferring it from received signal strength, which is the technique every proximity product before it had to rely on and which never really worked. It is the SIG's answer to the gap AirTag exposed. And Core 6.2 dropping the minimum connection interval from 7.5 ms to 375 microseconds is a twenty-fold cut in the protocol's latency floor, which matters if you have ever tried to push a firmware image across a link and watched an OTA update fight the connection dropping.

Where it ends up

The Bluetooth SIG's 2025 market update forecasts 5.3 billion Bluetooth devices shipping that year, approaching eight billion by 2029. The interesting number is not the total, it is the split.

Read that as a narrative rather than a table. Classic-only radios, the direct descendants of the 1994 cable replacement, are shrinking at fourteen percent a year. The low-power thing Nokia built to sit alongside them is growing at twenty-two. The province that got annexed is now most of the kingdom.

Bluetooth Low Energy is not a version of Bluetooth. It is a different radio that inherited the name, twice over.

None of this is trivia, incidentally. The reason a peripheral's advertising interval is the single most consequential number in its power budget is a decision made at Nokia Research before 2006. The reason an Android phone could scan from 2013 but could not advertise until 2014 is a platform choice with a date on it. When a BLE product behaves in a way that seems arbitrary, it is usually a twenty-year-old constraint still doing its job. If you are building on it and those constraints are biting, taking the firmware and the app together is usually the shortest way through them.