Re: Raspberry Pi 4's IP Address 169...

Yes, but I believe (and maybe I'm wrong) that the channels on 5 GHz are wider and so allow a greater data rate. Also there is a greater chance that the auto-channel-sensing logic in the router will be able to find a channel that is free of interference from neighbouring routers and from other sources of RFI,

Now that I didn't know. So maybe that explains why my old Win 7 laptop would not go much above that speed even when I was close to the router on an uninterfered-with channel with no other traffic on wifi. I thought the built-in wifi adaptor was failing, but maybe it was never designed to go much faster.

I built this place 20 years ago with full structured cabling - cat 5.

Ther is.

if for example your network is faster than the server disk I/O then loading a file off that server wont get faster with extra network speed.

I have yet to actually try gigabit over my wiring.

^^^^^^^^^^

Really? :-)

IIRC a Pi only does 802.11n natively and 150 Mbps and that's half duplex of course so will be less.

I think my Pi Zero W with is 5 feet away, through a wall from the wifi router does around 20-50Mbps depending on random reflections etc...

That's all on 2.4Ghz.

I dont have any devices bar my router than run on 5Ghz.

But I do have lots of cat 5!

The day Linus released Linux 1.0 I (and untold numbers of others) made the mistake of attempting to download it, contributing thereby to DDOSing Finland! Before I abandoned the attempt I got to see a very rarely triggered bug in my ftp program as it reported a download speed of "1 bytes per second".

Yes, really. From 0.5 Mbps with ADSL (the maximum that my router every synced) to 5 Mbps with VDSL. Made a very big improvement if I was sending emails with large attachments, sending them over TeamViewer or FTPing them.

Ah, I see what you were referring to: I did of course mean 5 Mbps, not 5 bps ;-)

Whoosh... Look closely at what you wrote, young man...

The nagging detail that my server disk I/O can be faster than the two bonded gigabit cables can carry is what occasionally tempts me to

Fast enough before everything got silly. I tried it out. I can read text at up to 300 baud and I can scan text, to find where to stop, at 1200 baud. Anything above that becomes a useless blur.

What silly like things that aren't text ?

Then again I used to think that sending a screenshot image instead of typing a bit of text was the extreme of lazy bandwidth waste - until I got sent a high definition video panning around a screen at close range from which I was expected to glean two small pieces of essential text - needless to say one was obscured by glare.

On Thu, 13 Jan 2022 09:36:56 -0000, "NY" snipped-for-privacy@privacy.invalid declaimed the following:

Peruse

Not bad, but with caeveats

They don't understand spread spectrum ...there is no such thing as 'channel width - what there is is adjacent channel crosstalk that diminishes as the channels get 'less adjacent',

Spread spectrum is destined for high availability and freedom from blocking by single frequencies in crowded spectrum space and when I was working next door to its secret development, it was for battlefield radios.

It ended up in mobile phones, and then wifi.

The ideas is that you 'smear' the information over a large section of spectrum, using a special secret code such that you cant kill the whole signal with a strong carrier.

Some implementations use frequency hopping as well - so the center channel frequency moves around again to throw off potential interference. Wifi doesn't though.

Pretty sure that some wifi implentations use two widely spaced channels to double the throughput

so each transmitter and receiver pair exchange data that is all over half a dozen channels at varying strengths at the same time. So you only get the signals belonging to you - everyone else's sound like noise. Leastways that is how it worked for the military and I think GSM. Wifi may in fact use one code per wifi point. And do the separation into wifi devices using coding at the digital level

Needless to say data rate is as always governed by Shannon - so the more other stuff is going on the slower the links will be.,

We are _far_ from information theoretic upper bounds on our links, just FYI. Coding and frequency-division schemes for mobile networks can also be a lot more complicated (like CDMA).

I dont think so at ALL. examination of stuff like ADSL shows we are within a factor of two of shannon limits

For a 22MHz wifi channel at say 30dB s/n, you are looking at around 660 Mbps shannon limit.

Its the same order of magnitude as the achievable maxima.

I thought that wifi did have a defined bandwidth depending on the comms speed (ie which 802.11 standard it supports). I've always understood that channels 1, 6 and 11 are guaranteed not to overlap (at least for the earlier standards) and therefore you get no further benefit from using channels spaced more widely than 1, 6, 11, but that you get progressively more degradation as you move the channels closer than 1, 6, 11.

Sounds good. Which wireless standards support it, as an alternative to channels that are spaced more closely than the bandwidth of any given comms link (as you have with 2.4 GHz). Whenever I've run wifi spectrum analysis software, it's always shown me specific channel number for each wifi network that is in range, and the number of channels either side that each network is using. Spread spectrum is completely different and as long as there are holes at irregular places in the spectrum, a spread spectrum will use it.

Well 'guaranteed not to overlap' is ArtStudent talk for 'show < -60dB adjacent channel crosstalk' and so on

Radio is intensely analogue. Spectra don't have sharp edges!

No, wifi is spread spectrum. If you like its as if the frequency at any given time is somewhere within 22Mhz of a given centre frequency, and is constantly moving around, or you could say its heavily modulated by a high frequency noise (a pseudorandom code) that its energy is spread across several channels . Thats why you need to de convolute it with the same pseudo random code in order to make sense of it.

The whole idea is to get away from any natural signals that might 'look' the same.

Oh heck, why have a dog and bark?

Here's a pretty decent write up

All the "wifi" and bluetooth standards are spread spectrum based.

More detail than you probably want to know:

TL;DR Wifi and bluetooth are spread spectrum using many very narrow channels spread around a fairly narrow space, this is why the effect of crowding APs too closely is a slowdown rather than a stop.