3 ms·

Too much power for small devices like earbuds or iot devices.

amelius 6y ago
That depends on how long you want to use them. For some the trade-off against sound quality, features and user-friendliness might be favorable.
rektide 6y ago
the shield tv's controller (2015) was wifi based, & could transit audio. it's not tiny, but it shows wifi was viable for a latency-sensitive battery-powered system five years ago. for many headsets i'd think it would be quite doable but yes i'm not sure we've scaled wifi down to earbud sized devices. and yes power consumption is afaik higher. it feels mainly like an adoption / standardization problem. we have chromecast speakers galore & wow is it nice having devices available over the network. alas there's no standardizations, no cooperation, no system for wifi to be useful. other than the very limited controlled & closed ones.
- iancmceachern 6y ago
  That controller had a 2600 mah battery for ~60 hours of operation. Much larger in power capacity and volume than what is suitable for tiny earbud like devices. https://batteriesglobal.com/products/nvidia-shield-battery?variant=32349283844174&currency=USD&utm_medium=product_sync&utm_source=google&utm_content=sag_organic&utm_campaign=sag_organic https://batteriesglobal.com/products/nvidia-shield-battery?v...
  - rektide 6y ago
    Ok so 260 mah for 6 hours? Using some 5 year not-that-fancy-at-the-time old technology (Ozmo2000)? That does far far far far more than send audio? That is much lower-latency than what we'd need to compete with Bluetooth (which radically increases wireless power needs)? For comparison, Google Buds are 120 mah & 5 hours. All we need is a 2x improvement in power, for a far less complex device, & we've had 5 years? You are making the case for me my man. This sounds more doable than I thought!! Miniaturization is indeed a challenge. Who does make really small wifi units? Who do they sell to? How much smaller do they need to be? I suspect it's simply never been a goal. And why would it be? There's no protocols that make wifi even potentially compatible with these sort of consumer device systems, no matter the size! Even with what we have today though, a wifi microphone or wifi gaming headset should be easily possible, no challenge at all, and could have far better quality & likely latency than what we expect from bluetooth. And I rather believe we could make smaller wifi if we had an interest in doing so. We are limited by our imaginations, our imaginations and our lack of standards/will.
    - iancmceachern 6y ago
      Not for specific applications like bluetooth headsets. There are application specific ICs that are made for the specific https://www.microchip.com/en-us/products/wireless-connectivity/bluetooth-low-energy/bluetooth-audio-and-voice-solutions# https://www.microchip.com/en-us/products/wireless-connectivi... Much like ASICS are a big step up in performanceand energy efficiency for crypto mining, they are in this application as well.
neolog 6y ago
Wi-LE (low energy) https://dl.acm.org/doi/10.1145/3365609.3365853 https://dl.acm.org/doi/10.1145/3365609.3365853
- amelius 6y ago
  That's interesting. > Surprisingly, Bluetooth devices require nearly three times as much energy to transmit a bit of data at the physical layer than WiFi devices.
sgtnoodle 6y ago
I imagine that power consumption is proportional to amount of data and inversely proportional to latency and data rate. To minimize latency, the transmitting device needs to send smaller packets more often, leading to poor efficiency due to overhead. The faster each packet can be sent, though, the less time the transmitter needs to be powered on. I suspect that the only difference in power consumption would be due to the modulations that wifi uses vs. bluetooth, i.e. wifi is able to cram more bits into the same amount of RF energy. The tricky bit with radio comms is that you absolutely have to be able to cope with data loss. The easiest, lowest latency way is to just encode the data in a way that allows it to degrade gracefully. The listener is aware of the data loss, but isn't bothered by it. This implies that there's a lot of redundant information in the data. That is at odds with compression, though, because compression is all about removing redundant information; a loss of any bit of data results in severe degradation. Radio links that send compressed streams of data need to either rely on acknowledgements with retransmission on data loss, or they need to add just enough redundant information for the receiver to reconstruct the original data stream even after expected packet loss. The algorithms that provide that "forward error correction" work by distributing redundant information across multiple packets, which inherently adds latency since the receiver needs multiple packets to decode (or recreate) the payload of single packets.
- iancmceachern 6y ago
  Not for specific applications like bluetooth headsets. There are application specific ICs that are made for the specific https://www.microchip.com/en-us/products/wireless-connectivity/bluetooth-low-energy/bluetooth-audio-and-voice-solutions# https://www.microchip.com/en-us/products/wireless-connectivi... Much like ASICS are a big step up in performanceand energy efficiency for crypto mining, they are in this application as well.
  - sgtnoodle 6y ago
    I'm not sure I follow what you mean. Of course an ASIC can be designed to maximize the efficiency of any given RF device. I personally have a BLE enabled medical device implanted in my chest with a 10+ year battery life. Based on a neat x-ray image I have, it has a single BGA IC, half a dozen ceramic capacitors, what looks like a crystal oscillator, and a couple discrete diodes. My point is that there's inherent energy required to transmit information via radio waves given a specific modulation and framing. Less data takes less energy than more data, and less framing takes less energy than more framing. For the same data rate (digital audio), lower latency requires more power than higher latency because there's more framing overhead. Additionally, because of how noisy the medium is, there's an inherent need for redundancy in the transmitted data, and that redundancy must either be distributed in time or in space. Distributing it in time generally requires less power but inherently adds latency. Distributing it in space means transmitting the signal with more inherent power. That's true whether you use bluetooth or wifi and have an optimized ASIC. The difference between the two is the available modulations and framing. If wifi supports a modulation that's more inherently efficient at transmitting low latency audio data than bluetooth, then it would make sense to implement it. WiFi seemingly isn't designed to do that, though. It's designed to be good at transferring data at orders of magnitude higher data rates, using proportionately higher power. The corresponding framing is designed around that assumption. Legacy bluetooth is terrible, but so is legacy wifi. Presumably, the latest bluetooth standards make use of modern modulations and framing techniques to minimize power consumption and latency appropriate for its use cases. If wifi were strictly better for the same use cases, then the next bluetooth standard would simply be "bluetooth" branding on top of something like wifi-direct.

That depends on how long you want to use them. For some the trade-off against sound quality, features and user-friendliness might be favorable.

rektide 6y ago

the shield tv's controller (2015) was wifi based, & could transit audio. it's not tiny, but it shows wifi was viable for a latency-sensitive battery-powered system five years ago. for many headsets i'd think it would be quite doable but yes i'm not sure we've scaled wifi down to earbud sized devices. and yes power consumption is afaik higher. it feels mainly like an adoption / standardization problem. we have chromecast speakers galore & wow is it nice having devices available over the network. alas there's no standardizations, no cooperation, no system for wifi to be useful. other than the very limited controlled & closed ones.

iancmceachern 6y ago
That controller had a 2600 mah battery for ~60 hours of operation. Much larger in power capacity and volume than what is suitable for tiny earbud like devices. https://batteriesglobal.com/products/nvidia-shield-battery?variant=32349283844174&currency=USD&utm_medium=product_sync&utm_source=google&utm_content=sag_organic&utm_campaign=sag_organic https://batteriesglobal.com/products/nvidia-shield-battery?v...
- rektide 6y ago
  Ok so 260 mah for 6 hours? Using some 5 year not-that-fancy-at-the-time old technology (Ozmo2000)? That does far far far far more than send audio? That is much lower-latency than what we'd need to compete with Bluetooth (which radically increases wireless power needs)? For comparison, Google Buds are 120 mah & 5 hours. All we need is a 2x improvement in power, for a far less complex device, & we've had 5 years? You are making the case for me my man. This sounds more doable than I thought!! Miniaturization is indeed a challenge. Who does make really small wifi units? Who do they sell to? How much smaller do they need to be? I suspect it's simply never been a goal. And why would it be? There's no protocols that make wifi even potentially compatible with these sort of consumer device systems, no matter the size! Even with what we have today though, a wifi microphone or wifi gaming headset should be easily possible, no challenge at all, and could have far better quality & likely latency than what we expect from bluetooth. And I rather believe we could make smaller wifi if we had an interest in doing so. We are limited by our imaginations, our imaginations and our lack of standards/will.
  - iancmceachern 6y ago
    Not for specific applications like bluetooth headsets. There are application specific ICs that are made for the specific https://www.microchip.com/en-us/products/wireless-connectivity/bluetooth-low-energy/bluetooth-audio-and-voice-solutions# https://www.microchip.com/en-us/products/wireless-connectivi... Much like ASICS are a big step up in performanceand energy efficiency for crypto mining, they are in this application as well.

neolog 6y ago

Wi-LE (low energy) https://dl.acm.org/doi/10.1145/3365609.3365853 https://dl.acm.org/doi/10.1145/3365609.3365853

amelius 6y ago
That's interesting. > Surprisingly, Bluetooth devices require nearly three times as much energy to transmit a bit of data at the physical layer than WiFi devices.

sgtnoodle 6y ago

I imagine that power consumption is proportional to amount of data and inversely proportional to latency and data rate. To minimize latency, the transmitting device needs to send smaller packets more often, leading to poor efficiency due to overhead. The faster each packet can be sent, though, the less time the transmitter needs to be powered on. I suspect that the only difference in power consumption would be due to the modulations that wifi uses vs. bluetooth, i.e. wifi is able to cram more bits into the same amount of RF energy. The tricky bit with radio comms is that you absolutely have to be able to cope with data loss. The easiest, lowest latency way is to just encode the data in a way that allows it to degrade gracefully. The listener is aware of the data loss, but isn't bothered by it. This implies that there's a lot of redundant information in the data. That is at odds with compression, though, because compression is all about removing redundant information; a loss of any bit of data results in severe degradation. Radio links that send compressed streams of data need to either rely on acknowledgements with retransmission on data loss, or they need to add just enough redundant information for the receiver to reconstruct the original data stream even after expected packet loss. The algorithms that provide that "forward error correction" work by distributing redundant information across multiple packets, which inherently adds latency since the receiver needs multiple packets to decode (or recreate) the payload of single packets.

iancmceachern 6y ago
Not for specific applications like bluetooth headsets. There are application specific ICs that are made for the specific https://www.microchip.com/en-us/products/wireless-connectivity/bluetooth-low-energy/bluetooth-audio-and-voice-solutions# https://www.microchip.com/en-us/products/wireless-connectivi... Much like ASICS are a big step up in performanceand energy efficiency for crypto mining, they are in this application as well.
- sgtnoodle 6y ago
  I'm not sure I follow what you mean. Of course an ASIC can be designed to maximize the efficiency of any given RF device. I personally have a BLE enabled medical device implanted in my chest with a 10+ year battery life. Based on a neat x-ray image I have, it has a single BGA IC, half a dozen ceramic capacitors, what looks like a crystal oscillator, and a couple discrete diodes. My point is that there's inherent energy required to transmit information via radio waves given a specific modulation and framing. Less data takes less energy than more data, and less framing takes less energy than more framing. For the same data rate (digital audio), lower latency requires more power than higher latency because there's more framing overhead. Additionally, because of how noisy the medium is, there's an inherent need for redundancy in the transmitted data, and that redundancy must either be distributed in time or in space. Distributing it in time generally requires less power but inherently adds latency. Distributing it in space means transmitting the signal with more inherent power. That's true whether you use bluetooth or wifi and have an optimized ASIC. The difference between the two is the available modulations and framing. If wifi supports a modulation that's more inherently efficient at transmitting low latency audio data than bluetooth, then it would make sense to implement it. WiFi seemingly isn't designed to do that, though. It's designed to be good at transferring data at orders of magnitude higher data rates, using proportionately higher power. The corresponding framing is designed around that assumption. Legacy bluetooth is terrible, but so is legacy wifi. Presumably, the latest bluetooth standards make use of modern modulations and framing techniques to minimize power consumption and latency appropriate for its use cases. If wifi were strictly better for the same use cases, then the next bluetooth standard would simply be "bluetooth" branding on top of something like wifi-direct.