The ePMP1000' latency

Dear everyone,

Please help me get the information about the ePMP1000' latency, I  could not found it on specs.

Thank you so much!

It is on the spec sheet under performance. 17ms round trip. 

It really depends on what mode (TDD, TDD-PTP, ePTP) and what frame type you use... 2.5ms, 5ms, and flexible.

ePTP has lowest latency under good RF conditions of a 2-3ms, followed by TDD/TDD-PTP using flexible frames 8ms, followed by 2.5ms frames 10-15ms, and lastly 5ms frames 20-25ms. All of these are approximations and can very depending on load, number of subscribers, etc.

I think 17ms is a good average across the board.

3 Likes

Hi. It depends if you choose to use them in GPS Sync mode or not. GPS Sync has some awesome benefits in allowing frequency reuse, reducing guard bands, improving interference rejection, etc. BUT, it does that by scheduling and syncing it’s TX /RX frames, which means that using sync mode will add a few ms of latency. If you choose ‘flexible’ mode, then the latency is.lower, but it’s TX/RX is less ‘organized’ in a way. So, each mode has its benefits, and each has an impact on latency.

Dear,

This value is too big, normally, it is about 5ms.

Why?

Thank you!

this value is to big ?     to which are you referring too?  

 the GPS sync mode is the highest latency as eric said, 20 to 25 MS.   thats less than 1 frame in a 30 FPS video game for a subscriber.  

or wimax frame of 50 MS.      or a TDD LTE frame of 25 MS.   

all time synced system's have latency delays.      as everyone pointed out, you can run the in flex mode, or 2.5ms frame mode and bring that down to 8 to 10.   you can go further and run them in eptp mode and go down to just 2 or 3MS.  you can also run them in standard wifi mode.  and you'll probably see the 1ms ping you have seen with other gear.    any device runing in wifi mode CSMA you'll have major variation in latency. all true TDMA base systems will have more stable and predictable behavior under load.     of any product out there, I'd say the EPMP offers you the most flexibly in operating modes to meet your goal with the reasonable price tag. 

1 Like

@vantienphap wrote:

This value is too big, normally, it is about 5ms. Why?


Hi. The value of the latency is basically up to you - depending which mode and what features you choose. If the lowest latency is your primary concern, then you can choose a mode that is suited for low latency and have 2ms latency if you wish. You see, you can use these radios in a whole bunch of different modes, depending on your needs.

So, for example,if choose to use them in normal Wifi mode, they will have very low latency like most WiFi devices do. BUT, WiFi devices have a whole bunch of drawbacks.. It's not very resistant to noise, it doesn't play well with competitive interference, it actually causes self-interference with your own other Access Points, it doesn't really have any sort of scheduler, so all the clients can try to talk to the AP whenever they choose and stomp on each other, it has a near-far issue, it has hidden-node issues, it's not very scalable, etc, etc, etc. The results of all of these WiFi drawbacks is that performance usually goes down and latency usually goes up when you start to add more clients. So, using them in WiFi mode (and other WiFi based products) may advertise ''as low as 1-2 ms latency'', and that may be true with only a few client, but that latency may actually be 50 or 150 ms when you have 20 clients on the AP.

You can instead choose to use Cambium radios in ''flexible frame mode'' - which organised everyone into 'frames' and it is much better managed and schedule, and still has fairly low latency. It is a trade-off though - so you're trading a bunch of pluses like better organization and scheduling and scalability, for a little bit of extra latency. If you think about it, if the Access Point's scheduler hands out a ''schedule'' to 20 clients and tells each of them when they get their turn to talk - then client 14 might be told to wait 2.4ms before it transmits it's information back to the AP. This is a good thing - it means that all other clients will be quiet at that instant, and it'll mean that the AP is ready to receive at that instant, BUT it also means that there is 2.4ms of extra latency introduced as a result. This is the way ALL properly scheduled carrier grade equipment works - a proper TDMA scheduler organizes everyone better, but it organizes everyone into a lineup sort of, and the last guy in the lineup is.. well... last. In 'flexible frame mode' from my home to the shop, I average 9ms latency. So for me, going from 2ms to 9ms average is a great trade off - it TDMA mode works so very well, that a few extra ms of latency is more than made up by the extra performance. (Attached, a screen shot of my latency in 'flexible frame mode'.

You can also choose ePTP mode (if this is a Point to Point link) and that mode has only a couple ms latency. It is again a different mode, with different pros and cons, and it's VERY low latency and its very high throughput.

You can also choose a GPS Sync Point-to-Multipoint mode with a scheduled duty cycle - so for example you can tell the Access Point that you want to reserve 75% of the time for data TXing to the client, and 25% of the time for data RXing from the clients. This is another mode, and this feature (again) adds an extra level of scheduling and allows rather wonderful benefits, like the potential for frequency reuse, and the ability for your own nearby Access Point to limit or eliminate their own self-interference. Since all your Access Points are now scheduled to the same GPS source, and since they are all transmitting and all receiving for the same ratio of each frame, things are all organized and synchronized across the whole network. BUT that scheduler adds in a few more ms of latency as a result. Again, that's not a bad thing - it's a very, very good thing. Yes, it is a few ms of latency, but as others have pointed out, it is dependable, scalable latency and the trade-off is this suite of benefits in performance, interference mitigation, and scalability.

1 Like

Thank everyone so much.