Why does wifi sometime do strange things?

Discussion in 'Networking, Telephony & Internet' started by 999Warrior, Feb 7, 2020.

  1. 999Warrior

    999Warrior Member

    Joined:
    Jan 25, 2019
    Messages:
    54
    The vast majority of the time I find wired ethernet 100% stable.

    All my devices are able to detect each other and Windows file sharing works.

    However over the years with wifi I've found at various points with different routers, switches and access points things go undetected, e.g. chromecasts, windows file sharing

    and strange things make them start working e.g. access DLNA via my Playstation and suddenly the whole network works over wifi not just wired ethernet.

    Why does wifi do these weird things sometimes?
     
  2. Hive

    Hive Member

    Joined:
    Jul 8, 2010
    Messages:
    5,871
    Location:
    Some place far away
    You get what you pay for really... cheap low end WiFi gear is abysmal
     
  3. OP
    OP
    999Warrior

    999Warrior Member

    Joined:
    Jan 25, 2019
    Messages:
    54
    Yes that is certainly something I've noticed but usually you don't know why its not doing something that the more expensive stuff seems to do fine.

    The only exception to this is in my experience has been TP-Link gear. I've always found it seems to work pretty rock solid despite the price of products you buy.

    Usually however, I just stick with cables as much as possible simply because its cheap and works for most stuff.
     
    l_ QuadX_l likes this.
  4. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    40,426
    Location:
    Brisbane
    Switched Ethernet: two devices having a relatively private, 1:1 discussion

    Hub / WiFi: many devices standing in a room all screaming over each other.

    Consider the efficiencies of each. Add in the extreme complexity of wireless modulation and environmental signal interference. Now you have your answer.
     
    Zee likes this.
  5. OP
    OP
    999Warrior

    999Warrior Member

    Joined:
    Jan 25, 2019
    Messages:
    54
    I guess this explains why when I use ac wifi with only 1 laptop on a specific SSID that is seperate from my router works pretty much 100% where as with the same SSID as my router it would bounce between them sometimes and sometimes I couldn't chromecast.

    It's almost identical to a 1:1 wired connection. 1 device going to 1 access point (being a reused old telstra router from after my grandparents died) with minimal interference cause very few others are using 5ghz n/AC wifi most of the time.
     
  6. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    40,426
    Location:
    Brisbane
    Yeah, almost identical, except for all the environmental noise and layer 1 latency inherent to wireless communication and information propagating over a non-conductor.

    So... I guess not identical at all.
     
  7. caspian

    caspian Member

    Joined:
    Mar 11, 2002
    Messages:
    10,725
    Location:
    Melbourne
    from layer 2 up, sure. pity about the layer 1 environment.
     
  8. hawpinghaxbag

    hawpinghaxbag Member

    Joined:
    Jun 6, 2005
    Messages:
    5,399
    Location:
    Your mum's place
    The air interface is constantly improving, 802.11ax looks pretty good but thats probably a part of the problem, constantly changing network standards
     
  9. Quadbox

    Quadbox Member

    Joined:
    Jun 27, 2001
    Messages:
    6,338
    Location:
    Brisbane
    Sure, but the same can be said about twisted pair ethernet. Twisted pair is a fucking horrible interface, we only use it because it's cheap and available and doesnt involve someone performing surgery on coax (in that you cannot do insulation displacement connections on coax), or on meticulously polishing the ends of a piece of glass optical fibre cleaved carefully to the right angle, and then checking it for imperfections under a microscope. And I'm saying that as someone who routinely needs to bloody look at fibre under a fibre scope, FML. Every single generation we've gone up an order of magnitude in transmission speed, the engineers of the various bodies who participate in the IEEE have literally thrown their hands up in the air on twisted pair, and said "no fuck it, lets just not", and eventually have cobbled together a relatively bad solution using clever encoding and so on to make it happen. But the result has been the same each time, for each generation since gigabit ethernet, the twisted pair version has been QUANTITATIVELY WORSE than the fibre and other alternatives, in every single way but cost.
    The very same thing applies to wireless, but the issues are orders of magnitude worse. High bandwidth RF communications free of a waveguide *fucking suck*. We spend vast amounts of time trying to find workarounds to this fundamental problem. But there is simply no way for wifi to be all three of these things simultaneously: cheap, reliable, fast. those three things being measured to the wired standard of the day, at any given time. Anyone who thinks otherwise is pulling shit out of their arse: the same noise cancellation techniques that make wifi work at high speeds, or reliably, or fast, apply *MORE* to copper connections, in exactly the same way. As such, for any given generation of tech, generally speaking your wireless tech is always going to suck more. Oh and boy are we testing the ends of this ancilliary version of moore's law. Even quantum radar equivalents in communication are only going to get us another order of magnitude if we're lucky, all that might help us with is reducing the SNR, which for closed fibre loops arent the limiting factor usually, dispersion and so on are. open space RF? lol, no.
     
    Zee and elvis like this.
  10. Quadbox

    Quadbox Member

    Joined:
    Jun 27, 2001
    Messages:
    6,338
    Location:
    Brisbane
    And to go further on this, many of the new protocols are actually going to produce licensing problems for ACMA. the ISM band is largely licensed under the idea that each individual transmitter's not going to be arbitrarily saturating it. Of course noise in the ISM band is nothing new, but there are a whole bunch of devices that assume there'll be some clear space *SOMEWHERE* on it, and some of those devices are a hell of a lot more important than your home wifi. ACMA and other international agencies of similar intent are probably in the end going to have to rethink their entire spectrum allocation to some degree, and the same applies to the .11ad and ax band. This isnt news of course, there's a whole bunch of emerging stuff that's probably got acma at present scratching their heads and saying well fuck, like starlink, oneweb and their like.
     
    elvis likes this.
  11. hawpinghaxbag

    hawpinghaxbag Member

    Joined:
    Jun 6, 2005
    Messages:
    5,399
    Location:
    Your mum's place
    I didnt realise they were 60 GHz, thats insane. I was thinking more about things like power control, duplex and multiple access techniques to improve the connection etc but i suppose they need to cater for cheap and basic devices at the same time
     
  12. Sledge

    Sledge Member

    Joined:
    Aug 22, 2002
    Messages:
    8,059
    Location:
    Adelaide
    Isn't that a well known fact for almost everything though?
     
    Pugs likes this.
  13. Quadbox

    Quadbox Member

    Joined:
    Jun 27, 2001
    Messages:
    6,338
    Location:
    Brisbane
    Well quite.
     
    Pugs likes this.
  14. elvis

    elvis Old school old fool

    Joined:
    Jun 27, 2001
    Messages:
    40,426
    Location:
    Brisbane
    Sure. But WiFi is one of those insanely complex things we take for granted every day.

    Compare what's going on at a pure maths level there. Like, hit up a few Wikipedia articles on wireless comms and see how long you last before your brain just taps out.

    The fact that we just casually bitch about it is kind of amusing.

     
    Sledge likes this.
  15. OP
    OP
    999Warrior

    999Warrior Member

    Joined:
    Jan 25, 2019
    Messages:
    54
    Beyond Cat6a, what about plastic optical fibre like this PVC fibre cable?https://www.fs.com/au/products/4020...pCQ-FTaBBj8SX29BG4kJwiYR3_UAAiYIaAt41EALw_wcB $19 for 30m of fibre.

    Just need SFP ports/converters to come down in price.

    Ah the beauty of wireless and clutter free.

    Downside is that if you invest hundreds of dollars in the gear to do it well you could pay for an electrician with a cabling licence to come in and cable up your house and it would last a lot longer and stability would be guaranteed.

    Still no cables sounds nice.
     
  16. Quadbox

    Quadbox Member

    Joined:
    Jun 27, 2001
    Messages:
    6,338
    Location:
    Brisbane
    Yeah I was more speaking to the miracles of the ever-in-the-future next big thing of networking, rather than gigabit speed stuff. Gigabit ethernet is, afterall, over twenty years old, and all those technical niggles I mentioned have long since happened. It's also not particularly high bit rate, we were still well within the realms of what twisted pair was capable of being good at, at that point. Currently it's *just barely possible* to achieve 40GBE in a lab, over 8 conductor twisted pair. Shielded twisted pair that is, not UTP. Whereas we've been doing 100GBE over single mode fibre for a loooongish time now, even without DWDM. And once you throw in wavelength division multiplexing on fibre, any comparison goes way out the window.

    EDIT - but back to the original point, that wifi is just now, twenty years later, starting to get to the point where it's vaguely close to equivalent to plain old 1000base-t
     
    Last edited: Feb 9, 2020
  17. cvidler

    cvidler Member

    Joined:
    Jun 29, 2001
    Messages:
    13,091
    Location:
    Canberra
    there's room for both style of connectivity, indeed you should be using both where each if best suited.
    If it moves regularly = wireless, if it's fixed = cable. best of both worlds.


    and that fibre cable you linked in still glass,it's the jacket that's PVC. only plastic fibre standard I know of is TOSLINK (optical digital audio).
     
  18. scips

    scips Member

    Joined:
    Apr 10, 2004
    Messages:
    469
    Look at the state of most network cabling in homes and businesses (in wall, racks, around the office).

    Fiber doesnt last 2 days if users can touch it, first time they pull it out of the SFP and check if anything is wrong? dead. Step on it with some deep tread shoes on carpet? dead. Pull it when its caught down the back of the desk? dead. Lick the end like a 9v battery? maybe cut tongue but cable ends would need to be cleaned.

    Copper is tough AF in comparison, I've used a cable as a dog leash (10m long) and it tests fine.

    Its gonna be around for a long time yet, especially in this backwards country.
     
  19. cvidler

    cvidler Member

    Joined:
    Jun 29, 2001
    Messages:
    13,091
    Location:
    Canberra
    fibre is NOT that fragile.

    more fragile than utp, sure. you could use the utp to tie up the office furniture to the roof. But fibre patch leads are more than up to day to day rigours of office use.

    Dept. Defence has fibre to every workstation/printer, indeed anything outside of a rack* (well over 100,000 end devices). they're not replacing fibre path leads every day as you'd suggest.



    *their security policies demand fibre everywhere, unless it remains within a secure rack, even between racks in the same row/DC must be fibre.
     
  20. evilasdeath

    evilasdeath Member

    Joined:
    Jul 24, 2004
    Messages:
    4,890
    Yeah the ISM bands are going to be a problem always for wifi. The problem is that its not just wifi, microwave ovens, remote control cars, nanny cams, some of which don't actually care about using the spectrum well, just that they do, and everyone tries to overcome everyone else. IE i need to reliably send 2 bits of data that must be correct, let me just send it 100 times and saturate the channels, great for me crud for everyone else. The other problem is 5 bars signal, everyone wants it, which means they turn there wifi up full ball, buy the range extenders, or the extended range wifi access points, and if one manufacturer designed things to not be so loud all the time, everyone else would just take advantage of it.

    You haven't worked with much fibre have you, thats all bullshit, fibre patch leads are tough as. Or Lab cables get treated like crap and they still work. A fibre patch lead is honestly 99% protective, only when you get down to a single strand does it get easy to break. I'd happy lick the end of a fibre patch lead, and it would still work, may not to full range but it would still work, and wouldn't cut my tongue either, it has a protective ferrell around it. The only reason not to plug it in is to stop depositing crap on the inside of the optic, which is easy enough to clean these days anyway. I have seen and damaged both.


    As for the Ops problem, the issue is less with wifi itself, it's more with multicast which most consumer devices do a piss poor job of. The protocols use it for discovery.
     

Share This Page

Advertisement: