Hacker News Reader

Comment by oflannabhra | Hacker News Reader

The best presentation I've seen on BLE throughput is a 2017 WWDC session on CoreBluetooth [0][1].

Upping the MTU is "easy" on Android (you can do it, but I've had mixed results), but requires HID on iOS. In fact, you cannot trigger an MTU negotiation through CoreBluetooth, so that a single peripheral cannot take over the entire radio. If you control the peripheral firmware, you can obviously request an increase from that end.

The best checklist I've seen is in that presentation:

1. Use Write Without Response

2. Chunk your data based on the connection interval, so you maximize # of packets sent per interval

3. Negotiate a larger MTU

4. Use BLE 4.2 Extended Data Length extensions

[0] - https://asciiwwdc.com/2017/sessions/712?q=core%20bluetooth [1] - https://devstreaming-cdn.apple.com/videos/wwdc/2017/712jqzhs...

View on HN

Thanks for sharing the link! I remember watching that presentation ... I really liked how thorough it was. I'll include the links in the article as an additional reference!

> Upping the MTU is "easy" on Android (you can do it, but I've had mixed results)

hehe, yeah, I've definitely run into some interesting issues when changing the MTU on android. There's a class of older devices that seem to only support the default MTU size but will happily negotiate a 512 byte MTU and then drop the data for you :)

Your article is great, too! I've saved it along with other Bluetooth resources for reference. Way more accessible than a video, thanks for posting it. There was nothing like it when I was deep in needing to maximize throughput.

The thing I really appreciated in the presentation was the visualization of how each of the options increased throughput. I also was very thankful that CoreBluetooth handles so much for us. I was able to get our firmware updates to get to 85% theoretical speeds without having to do much fiddling (besides good algorithm design).

I am jealous of your lab gear, though. I got Ubertooth [0] running for sniffing, but could only successfully follow a connection event about 25-33% of the time.

[0] - https://github.com/greatscottgadgets/ubertooth

> I also was very thankful that CoreBluetooth handles so much for us. I was able to get our firmware updates to get to 85% theoretical speeds

That's awesome! I've generally been pretty happy with the CoreBluetooth framework as well and it's nice that all devices in the iOS ecosystem behave fairly consistently. I have found Bluetooth State Restoration to have some odd quirks from release to release but I think it has gotten better over the years as well.

> I got Ubertooth [0] running for sniffing, but could only successfully follow a connection event about 25-33% of the time.

heh, yeah, the Ubertooth is neat for the price point ... I've used it as cheap spectrum analyzer [0] several times in the past when prepping for BLE RF testing. Could pretty quickly verify that a bluetooth chip was at least entering Direct Test Mode and transmitting on a particular channel using it!

[0] http://ubertooth.sourceforge.net/usage/start/

That's a great presentation! I had not seen it. Thanks for sharing.

Reply on news.ycombinator.com