Hi guys, I need some suggestion about a very special configuration:
point to point link;
distance only 1 meter (cannot use cables, only wireless, these are detachable and moving objects);
(true) full-duplex;
low latency,
throughput is not important, but reaching high troughput is a plus so AC radio instead N could help;
redundancy is a plus too but at this stage we can skip it;
very stable and reliable;
cost is not so much important, so any suitable MikroTik product is fine: However this is just a test, so I don’t need special features, outdoot box, because this test will run in lab enviroment.
How I will use such configuration ?
I’ll built a test in a lab with 11 devices so to have wireless 10 hops;
Each hops between wireless device is continued with an ethernet cable or a switch;
The first and the last device will be connected to the testing network (where will be attached a networkg analyzer)
hAP lite are capable of trasmitting full duplex ?
Shuold I choose a device with dual radio to get a true full-duplex point to point setup ? (using protocolos live nv2 or nstreme2)
5.7-5.8 GHz band could be better for such short distance ?
Hap lite is dual chain. You can use one for tx and second for rx only. If you need. Maybe the results will be so good for you (you stated no need for high throughput) that you even will not have to split the chains. For such small distance there will be probably no difference between the two frequency ranges. And the price matters, no?
Dual chain is good. But it seems it has embedded antennas, while I need to have esternal plug to attach some attenuator because I’ll conduct some test in lab enviroment.
I need to add more explanation points:
I’ll built a test in a lab with 11 devices so to have wireless 10 hops;
Each hops between wireless device is continued with an ethernet cable or a switch;
The first and the last device will be connected to the testing network (where will be attached a networkg analyzer)
Attenuator will be choosen to emulate such short distance and eventually to reduce/calibrate the trasmitting power.
I want to measure the end-to-end latency and throuput when trasmission is full duplex.
Throughput is not important, but reaching high troughput is a plus so AC radio instead N could help.
Redundancy is a plus too (so two indipendent radio card or two devices for each hop behind a switch are fine), but at this stage we can skip it.
Cost is not so much important, so any suitable MikroTik product is fine. However this is just a test, so I don’t need special features, outdoot box, because this test will run in lab enviroment.
Looks like so specific project that you personally would need to make your tests with the devices you believe will be the best. Generally, because of the low gain internal antennas of hap lite and ability to reduce the tx power to zero, you would not need any rf cables and attenuators between the radios. But it is just my opinion, you know the best what exactly you are going to build…
Do not expect that nv2 (or any other tdma protocol) will lower the delays. In case of low throuhgput the latency will be lowest with 802.11 protocol. But you will see during your tests.
Keep us updated with your results and provide some pictures if possible. It can help to others.
Yes it’s a very specific (industrial) project.
I would like to use attenuators because in that way I’ve not a real spatial stream transmission, and I can put each device near to the other, on a table. Otherwise I’ll catch interference of other devices (hop) and I’ll lost the control of the simulation. Inside a lab atteanutors are very useful to simulate the distance attenuation, cabling devices on a table. Outside the lab the distance will be different (about 25 meter each couple from another, but inside iron isolated floating cabins).
Regarding nv2 or similar, the main requisite is to establish a full duplex transmission preferibilly on different channels, so not to diminuishing the bandwidth in the end-to-end communications (so between the first and the last device of the wireless chains); with normal Wi-Fi transmission the available bandwitdh will be half at each hop, just like when using WDS protocol, so I’ll the end-to-end bandwidth after 10 hops will be seriously diminuished.
If bandwidth will be “quite” preserved, we will check the latency.
If it will work in lab, we will able to reuse that setup, without attenuator, in the real enviroment. If it still will works good, it can be used in thousand of production enviroment.
So, three question:
OSPF could be an alternative for using different radio links for trasmitting and receiving in full-duplex, at each hop, so from end-to-end will be like using an ethernet cable. What do you think about it ?
Then is better to have 1 radio with 2 chains or two radio modules for each devices ?
An alternative to hAP lite or hAP AC but provided of connectors ? (we can consider to open the hAP and hack it, of course, but we prefer the simplest path )
Well,
generally, wifi connection is not full duplex at all. And it does not matter what protocol you use. If you insist on latency, it is the time from issuing a packet from point A to point B, reciving, processing and sending the relevant response back from B to A. When using 802.11 and not loaded link it could be under 1ms. With nv2 you will neet to wait to transmitting time windows so it will not be below 2ms, roughly. But as I said, you can easily try.
When you insist on full duplex link it looks for me that traffic in both ways will not be so much relevant to each other so maybe there is not so much pressure on the latency. But hard to say, the purpose is still not clear enough for me. But if you think the latency is crucial, it is not smart to use ospf as it takes some time to redistribute the routes in the network when the structure changes. That would definitely cause big delays. Using bridged network could give you faster response times when the client reconnects to other ap and will be much simplier. You should also share the same frequency between all involved devices to speed up the reconnections. Try the Capsman too.
Two radios are always better than one. Only by this you can achieve real duplex. Just be sure you have enough frequency space between the both links to prevent self interference. Check the nstreme dual setup in documentation.
If you need connectors then look at basebox, where you can put additional radio to have two radios for duplex link. rb912, rb922, rb433 and similar boards could be fine for you also.
