Hacker News Reader

Comment by pzo | Hacker News Reader

> Wi-Fi 6's comparable maximum rate is 9.6Gbps. There is no consumer problem of today, no domestic use case imaginable, where 50Gbps works where 10 will not.

iphone has just 2x2 MIMO antena so even with WiFi 6 can get only max 2.4Gbps (1.2Gbps per antena - someone correct me if I'm wrong). And those are only max speed in labs in perfect conditions. So there is definitely use case for WiFi 7 in consumer devices e.g. VR/AR/Drones streaming cameras, remote control etc. Higher bandwidth should also make connection more robust and overall latency possibly reduced.

I'm definitely looking forward for WiFi 7

View on HN

The 2.4Gbps phy rate is nowhere close to usable bandwidth. With wide channels, good reception, not too much interference, and no really old hardware interfering with your connection, you can get over 1gbps on it, but the theoretical maximum is irrelevant to most use cases.

In practice, most people don't have their routers set up for 160MHz bands (because older devices don't support it and will fall back to the much slower 2.4GHz otherwise) and in some areas near radars you can't even use those bands in the first place. However, with modern equipment, you can definitely get plenty of bandwidth over WiFi 6.

The 50Gbps number stated also assumes very wide channels and even more recent devices, if WiFi 6 isn't doing it for you today, WiFi 7 won't be that much better.

>Higher bandwidth should also make connection more robust

Nope. Higher bandwidth could mean two things:

Larger channel size: larger channels have a higher noise floor and are are therefore less robust

Faster throughput: Faster throughput generally means larger channels (as above) or various techniques to get more bits/Hz (such as more complex modulation), both of which are less robust

I was more thinking if there is some frame collision and have a lot of data to send then overall robustness latency should be reduced? Similar like in a bus even if they travel at the same speed if you miss your bus or bus is full you have to wait. On the other hand if the bus is bigger with more sits or bus scheduled more frequent it's less likely you will have to wait. Also if bus breaks there is another one not far away behind to pickup passengers from broken bus.

I think I read this somewhere but I am struggling to recall where, so hopefully I didn't dream this: I believe the larger channels are much, much larger and include adaptive techniques to dynamically exclude noisy regions within channel. It can afford to entirely sacrifice bands in this way because the channels are much larger.

Even in the absence of all anthropogenic RF noise, the noise floor on larger channels is higher. Specifically, I'm thinking of thermal noise: https://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise

What you're writing isn't really correct either, this partly stems from the difficulty of what we mean by "robust".

To increase throughput (i.e. bitrate ) we can either increase the channel bandwidth or the SNR (and use that SNR for higher modulation formats). Increasing the channel bandwidth does increase the noise power (assuming matched filtering) but we typically also assume that the psd of the signal stays the same => signal power increases and SNR stays the same.

If we mean by "robust" that the signal is not as susceptible to fluctuations of the noise, than increasing the bandwidth could help. Assuming we want the same bitrate, we could use the larger bandwidth to reduce the modulation format and thus the required SNR while keeping the bitrate the same. However a larger bandwidth typically also increases probabilities of impairments (interferers, filtering...) but this is typically still a win, because capacity throughput is linearly proportional to bandwidth but only logarithmically to SNR.

I'm honestly struggling to come up with a use case for a single iPhone to need more than 300 megabytes per second (2.4Gbps) of bandwidth at any given time. (Noting that 8K video, in fairly un-optimized scenarios, appears to need about 96Mbps—just 12 megabytes per second, or 4% of that "low" bandwidth.)

What are you doing, streaming raw, uncompressed video from multiple sources to a single iPhone? If you've got that many devices you need to stream from...it seems to me that it makes sense to invest in some dedicated hardware to handle that, rather than expect to do everything with just an iPhone.

for iPhones maybe not but for drones or AR devices if you want to stream 4k at 60fps or 120fps then there is a lot of data to send. You could use iPhone as Drone payload for doing some site/construction inspection etc. Same with drones - they can fly very fast so would be nice to stream at 4k@120fps. Drones or iphones can/have many camera including lidar, depthmap (trudepth)

That is only the case if you wanted to stream uncompressed data (which for drones is defeinitely not needed), so VR/AR maybe, anything else it would be a waste.

That is only the case if you wanted to stream uncompressed data (which for drones is definitely not needed), so VR/AR maybe, anything else it would be a waste.

compressing big video frames (4k) takes time. If your application is latency critical (and drones moving at 100km/h is) and you are streaming at 120fps you have only 8ms for video frame processing. Sometimes you also want uncompressed data (lidar, depthmap) not for human but for machine computer vision / processing.

...Then what was the point of specifically calling out the 2x2 MIMO antennae of the iPhone?

that 9.6Gbps speed mentioned in the article doesn't apply for most consumer devices because they have only 2 antennas. Only expensive routers have many antennas to reach such speed. Not sure if we can have 8 antennas in smartphones

Faster link speed allows the Wi-Fi radio to finish transmission faster and go to sleep prolonging battery life.

TCP throughput is about half of the link rate. iPhones before the 15 Pro maxed out at 1200 Mbps link rate because they are 2x2 MIMO with 80 MHz max channel banwdidth.

There may be diminishing returns on watts per bit, something that 10gbe Ethernet is not great at.

Downloading big games (apple showcased Resident Evil village I believe)

According to a quick DuckDuckGo search, RE Village's download size varies between about 26GB and 29GB depending on what platform it's for—let's just round to 30GB for convenience.

At that size, and the speed I mentioned above, it would take less than 2 minutes (100 seconds, to be precise) to download the entirety of the game. I don't know about you, but I think expecting that a game like that download in less time than that—an operation you only have to do once, when you first install it—seems pretty excessive.

And MIMO means the AP can talk to multiple such clients at the same time, even if each iPhone is only using 2x2.

Streaming raw video to VR headset is probably only situation when it makes sense for consumers. For near every single other thing it's a massive overkill

There's no qos or control over the radio channels. A 120gbit, 120fps, 8k stream is lovely, until you have network interference of 10ms and you drop a frame.

another example: Apple added camera continuity for latest tvOS, you might want make video call with 4k at 60fps - sure you probably want to compress it but still you want to reduce latency.

Also with smartphones that have 1 TB of storage would be nice to use it as fast wireless hard drive

Reply on news.ycombinator.com