Artwork with a very hi density network
-
1,000 wireless devices? No. Because physics.
-
Hi John, thanks for your quick reply.
Actually, I am not sure as I am not an expert in network.
I need one or several access points (definitly).
I need to protect my network (wep or wpa key)
I need the server and the devices to exchange udp packets
And I thought using mini pc (like raspi) or old PC's for the hardware and pfsense for the software.
But may be it is not what I need.
In which case, what do I need ?W4RH34D, I read somewhere that a football stadium would provide up to 25000 wifi connexion to their visitor.
What do you mean by impossible because physics ?Thanks again for your help to both of you.
-
They also have an entire stadium of people spread out over vast distances.
Wireless in a stadium is not a trivial feat, either.
It's easy to relate this to cellphones as it is very similar. Do you think 1,000 cellphones in the space of 100 square feet can all dial out at the same time?
What if everyone flushed the toilet at the same time in a city?
https://en.wikipedia.org/wiki/Radio_spectrum
https://en.wikipedia.org/wiki/Microwave_transmission
https://en.wikipedia.org/wiki/Path_loss
https://en.wikipedia.org/wiki/Radio_propagation_model
https://en.wikipedia.org/wiki/Channel_(communications)#Channel_models
https://en.wikipedia.org/wiki/MultiplexingYou would need some sort of short band wireless spectrum to coordinate all of the devices and one master device on a different spectrum acting as the uplink - in short - it would be like a very long line of devices taking turns like before mu-mimo - only with a large number of devices you would get a lot of lag from all the ques.
-
Yeah a 1,000 devices your going to need lots of AP. And have to work out method so different devices connect to specific AP. But with everything so close your going to run into interference.. 50 meters is a very small space..
As W4RH34D mentions doing wifi in stadium sort of settings is quite difficult. When was the last time you were in a large crowd and your cell phone actually worked. Kentucky derby for example - cell phone is useless ;)
-
This is a great discussion topic, actually.
Wi-fi calling acts kinda like what I'm talking about.
If you could somehow get NFC to coordinate devices on when which device is going to use a full wireless spectrum - it may be "laggy" but it would probably be better than nothing.
-
1000 is not that many. Good APs can handle a bunch. What you did not state is whether you are constrained to 2.4GHz or if you can use 5GHz.
You can almost certainly get enough APs (probably about 5 good ones) on 5GHz with no co-channel interference.
You will probably want to position them to utilize as much of the terrain as possible to try to make one AP the obvious choice to a slice of the area. In the stadium example above, they like to use concrete enclaves to separate the APs from each other. I would not be surprised if such things are not part of the design in new facilities. You will probably need something controller-based that will do whatever it can to balance the load among the different APs.
Doing high-density wi-fi is not cheap. Getting APs that do not fall on their face with 200-ish associations on one radio is not cheap.
-
1000 is not that many in a 50m area??
-
1000 is not that many in a 50m area??
Well 1000 devices of which one is watching youtube, it would be fine, right? ;D
-
It's a lot but with a handful of good controller-based APs (think Ruckus) that can deal with it and low bandwidth requirements I'll bet it's doable. Far easier on 5Ghz than 2.4. You'll want to make sure dot11b beacons are turned off, etc.
No latency requirements given either.
-
It's a lot but with a handful of good controller-based APs (think Ruckus) that can deal with it and low bandwidth requirements I'll bet it's doable. Far easier on 5Ghz that 2.4. You'll want to make sure dot11b beacons are turned off, etc.
No latency requirements given either.
With enough money and man power anything is doable -right. But as a one off my opinion is if you have to ask - it's probably not tenable unless you have a ton of money.
-
Gear for that could probably be rented.
Probably would want to separate the broadcast domains using multiple SSIDs.
Probably don't need a firewall. A router or L3 switch would be fine.
-
Gear for that could probably be rented.
We thought so too - we had some big wigs come in and the service is terrible (for physical reasons and distance from the tower). We couldn't get Verizon corporate to bring out a cow to save their lives. (no fat pipes ISP in the area, either.)
-
And turn on isolation so every broadcast from wifi devices isn't going back out the wifi - only those from the switch, etc. Assuming the devices only need to talk to something upstream and not each other.
-
On precision : the area is rather 100 to 150 m wide but I was considering having several AP's spread around. That is why I told that any device would not be farther than 50m from an AP.
Now, reading your replies, it seems that the fact that my device exchange only a few bytes not so often, does not count so much.
I thought it would help to increase the number of connection on a single AP.BTW, happy to have launch a so interresting discussion :-)
-
Well it's easy to set Tx strength in most AP's but not usually in consumer client devices.
You would first start I think at limiting the broadcast power to the space it needs to cover. 150m of clear, line of sight or are we talking 150m of doors, walls and wiring?
There is a shot - but depending on the complexity of the area you would benefit from some wireless analysis tools.
This also greatly depends like Derelict was saying on the consistency of protocol and wireless bands you are using.
A mixed G, N, and AC would be a lot more difficult then say all wireless AC wave2 clients and hosts - actually that would probably be pretty fun to use.
Do the client devices move or are they in static locations? There is a ton of variables depending on your requirements - Like I said - it would be somewhat of an undertaking for whatever result you are trying to accomplish.
https://en.wikipedia.org/wiki/Bluetooth
You could divide the load on wireless AC AP's by 7 if you configured them in groups of 7 to be bluetooth with one uplink host using 5ghz AC.
-
Wireless chipsets generally handle the number of associations they do independent of the amount of traffic.
IIRC on the ruckus gear things like supporting WEP or TKIP reduced the number of associations allowed. WPA2 with AES was the way to go.
Seems I have seen something like you are looking to do at the olympics.
Might see if you can find a writeup about what they did.
-
Wireless chipsets generally handle the number of associations they do independent of the amount of traffic.
IIRC on the ruckus gear things like supporting WEP or TKIP reduced the number of associations allowed. WPA2 with AES was the way to go.
Seems I have seen something like you are looking to do at the olympics.
Might see if you can find a writeup about what they did.
No Rube Goldberg machine, then?
http://blueballfixed.ytmnd.com -
Someone has far too much time on their hands.
-
Just enjoying the time I'm not spending banging my head on a desk before I moved to pfsense.
LOL
OP Are your client devices going to be made? Meaning do you get to choose the hardware/software in which to accomplish your goal or are you just going to try to use prosumer/consumer gear?
Do you require full duplex communication to each "node" individually? -
My devices are all going to be the same.
If possible all made with this very cheap and simple component : ESP8266 (https://en.wikipedia.org/wiki/ESP8266) which can be bought for less than 2$, is compatible with 802.11 b/g/n (only 2.4GHz) and have an integrated GPIO which can handle up to 16 I/O.
They will definitly all move (located on a hat worn by people) in an open area (no walls, no doors) …
And once again, all send and receive only a few bytes every 5 seconds with an acceptable delay of 1 sec.Regarding the price, I expect a price of 3 to 4$ per device => 3000 to 4000 $ total for the devices.
What would be the necessary budget for network infrastucture
To reduce the network infrastructure cost, I would prefer to build my own AP, using old PC's or mini PC's like banana Pi M3 (if that has any sense, sorry I am definitly not an expert in the network area)
