In this blog post we'll go through what Bluetooth HDT is and how much faster it is compared to Bluetooth LE 1M and 2M in the demo we’re running on a new prototype SoC from Nordic.

Introduction

Bluetooth High Data Throughput (HDT) is the next step in Bluetooth LE, bringing much higher data rates, and we’re developing an upcoming SoC that will be among the first to support it. In our demo, which you can watch later in this post, we’re showing the prototype in action for the first time.

Bluetooth HDT is almost finalized in the Bluetooth Special Interest Group (SIG) and anticipated to be released this year. It extends Bluetooth LE with new PHYs based on more advanced modulation and longer packets, designed to enable higher data rates than Bluetooth LE 1M and 2M can provide. It will sit alongside LE 1M and LE 2M and won't replace them.

This isn't the first time Bluetooth LE has become faster. Bluetooth Core Specification 5.0 added the LE 2M PHY, which doubled the symbol rate and kept the GFSK modulation and the packet structure as they were, so once both ends supported it, the same data went over the air in half the time. Bluetooth HDT works differently. It changes the modulation, the packets and the number of PHYs at the same time, and it comes with its own trade-offs, so it fits some applications very well and others less so.

A brief look back at LE 2M introduction

It's worth a quick look at what LE 2M changed when it was added alongside LE 1M, since we'll use it as the reference. The modulation stayed GFSK at one bit per symbol, but the symbol rate went from 1 to 2 million per second. The packet structure stayed almost the same too, and so did the link layer and the host.

The price for the higher throughput was slightly lower sensitivity than LE 1M, which means somewhat shorter range. Throughput didn't double, since the protocol stayed the same and the gaps between packets and the acknowledgements take the same time as with LE 1M. A connection can switch between the 1M and 2M PHYs at any time with a standard procedure, but in practice most products request LE 2M once and stay there.

What changes with Bluetooth HDT

Each of these four points looks different with Bluetooth HDT.

Longer packets and a new protocol

Where LE 2M reused the packet structure, Bluetooth HDT defines new protocols with longer packets, and they're more advanced and more complex than the ones LE 1M and 2M use.

The length matters because every packet has a fixed cost in time: the preamble, the header, the gap before the next packet and the acknowledgement. A short packet spends most of its time on that, while a long packet spreads it over a lot more payload, and that's where a large part of the throughput gain comes from. It also means that with short packets Bluetooth HDT has little to gain and can end up slower than LE 2M. The minimum payload at which Bluetooth HDT outperforms LE 2M is around 10 bytes, but this is a theoretical value that depends on the specifics of the application and environment. For high throughput, much larger payloads are recommended.

Higher throughput also means less time on air for the same payload, which leaves more room for everything else. The radio is free sooner for other traffic, such as other connections and other devices sharing the 2.4 GHz band.

PHYs with new modulation

LE 1M and 2M use GFSK, which carries one bit per symbol. Bluetooth HDT adds π/4-QPSK with two bits per symbol, 8-PSK with three and 16-QAM with four, combined with forward error correction. This is a different kind of modulation from GFSK, so it needs a new radio and can't be added to existing SoCs with software, which is why Bluetooth HDT will come with new SoCs.

Why so many PHYs, throughput vs range

The specification defines five new 2.4 GHz PHYs: HDT2, HDT3, HDT4, HDT6 and HDT7.5. Each HDT PHY is a different balance between speed and range: the fastest PHYs need the best link quality, while the slower ones with more error correction work efficiently over longer distances or in noisier environments. A link isn't constant. Distance, interference and the position of the user keep changing, and no single PHY is the best choice through all of it, which is why Bluetooth HDT has five.

With LE 2M, the PHY can be changed at any time, but in practice most products set it once and stay there. Bluetooth HDT is designed to switch the PHY instantly, reacting to the conditions of the moment, similar to Wi-Fi. A Link Quality feature in the protocol, handled on the Link Layer, gives the feedback a device needs to decide when to switch. That's one of the hidden gems of Bluetooth HDT, besides the higher maximum throughput.

See it in action

In our demo, two prototype SoCs exchange data over all the new HDT PHYs and over LE 1M and 2M, one PHY after another. Joel Kauppo presents it in the video below.

In this demo, we ran the same 2 MB file transfer over every PHY. On the fastest HDT PHY, the average throughput was about 3.8 Mbps, with a maximum of about 3.9 Mbps, and the transfer took about 4 seconds. LE 2M averaged and peaked at about 1.4 Mbps and took 12 seconds, while LE 1M averaged about 0.76 Mbps, peaking at about 0.77 Mbps, and took 22 seconds. The other HDT PHYs fill in the steps between, and on every PHY the average stays within about 4 percent of the maximum.

Choosing the right technology

Bluetooth LE 1M and 2M remain the go-to choice for ultra-low-power applications that don't need as much throughput. They use much shorter packets, with less overhead and lower latency. For these applications, the nRF54L Series is the right choice.

Bluetooth HDT will be suitable for the applications that need more. As a rule of thumb, the more data there is to move, and the longer the packets are, the more it gains over LE 2M, while for smaller, more frequent messages it gains less, or even performs worse below the break-even point.

Wi-Fi is where products turned when Bluetooth LE 1M and 2M weren't fast enough, and it came with a price: a second radio, more power, more design effort and a network to join. Bluetooth HDT will cover part of that territory: the transfers that are too much for LE 1M and 2M but don't need Wi-Fi's bandwidth. This will be possible with Bluetooth HDT-enabled Bluetooth LE SoCs, over the Bluetooth LE connection the product already has. Wi-Fi remains the choice where even more throughput is required.

What it unlocks

Bluetooth HDT suits products that need faster bulk data transfer or richer data streaming than Bluetooth LE 1M and 2M can offer. It's likely to fit best for data from a few MB to tens of MB, but it isn't limited to that range. Here's what that means for the end user:

Bluetooth features tend to find uses nobody planned for, and we're curious to see what engineers build with Bluetooth HDT.

What's next

Nordic will share more information about Bluetooth HDT-capable SoCs at a later date. In the meantime, if your application doesn't need this much throughput, the nRF54L Series is the right choice, and if you want to discuss Bluetooth HDT for your application, contact your Nordic Sales Representative. And keep an eye on the DevZone Blog, where we'll share more details as the adoption of Bluetooth HDT progresses. You can also share your email in this form, so we can reach out when we have news about Bluetooth HDT and our HDT-capable SoCs.