How important is synchronization in TR-398 Long Term Stability testing

Natiya

New Around Here
There are some procedures to measure long term stability behavior in routers such us sending traffic or doing pings from a couple of stations, for example, every 2 minutes.
These tests last about 24 hours, so I was wondering how relevant could be the delay that occurs for example in the ping starting time of two stations.
I noticed even if the ping duration is set to be the same in both stations, the total ping duration on each is not the same one, so each station will start the next ping at a different time. Would that be something to avoid by forcing the stations to send the pings/traffic at the same time on each iteration? Thanks!
 

ColinTaylor

Part of the Furniture
What "procedures" are you talking about? Can you be specific.
 

Natiya

New Around Here
What "procedures" are you talking about? Can you be specific.
Actually what I mean is sending a ping and traffic from each station periodically during the test. This is done in a loop manner throughout the tests so is it needed that traffic and pings start at the same time from the different stations?
For example, the duration of each ping is not exactly the same and there is a delay added after each iteration. This is the "natural" behavior but one can force the pings to start at the same time if needed. I'm not sure if this synchronization is something required or if it's not relevant as long as the connectivity is checked during the tests.
 

ColinTaylor

Part of the Furniture
What specific testing software are you referring to? Without know that we can't comment on how it's meant to be configured.
 

thiggins

Mr. Easy
Staff member
Ping measures round-trip delay between two endpoints, i.e. the time it takes the ICMP message to be sent and received by the machine originating the ping.

The time is measured relative to the machine originating the ping. So synchronization between the two machines involved in the ping is not necessary.

If you want to measure one-way delay, i.e. the time it takes an ICMP packet to travel from one machine to another, then the two must be time-synched. The latest version of iperf (aka iperf2) can make this measurement.
https://sourceforge.net/projects/iperf2/
 

Natiya

New Around Here
Ping measures round-trip delay between two endpoints, i.e. the time it takes the ICMP message to be sent and received by the machine originating the ping.

The time is measured relative to the machine originating the ping. So synchronization between the two machines involved in the ping is not necessary.

If you want to measure one-way delay, i.e. the time it takes an ICMP packet to travel from one machine to another, then the two must be time-synched. The latest version of iperf (aka iperf2) can make this measurement.
https://sourceforge.net/projects/iperf2/
Thank you! So in the test Long Term Stability of the TR-398 where two stations are sending traffic and pinging the AP in a loop manner, is the one-way delay something relevant or is it check to check connection losses so it doesn't matter the delay introduced after each iteration?
 

coxhaus

Part of the Furniture
I don't see this as being important. This seems like a way to meet minimum requirements to be out the door for sale.
 

Natiya

New Around Here
I don't see this as being important. This seems like a way to meet minimum requirements to be out the door for sale.
In the TR-398 says that ping is to count the number of ping failures, so I guess it's ok if both stations are pinging the AP all the time, even if the pings don't start at the same time.

However, I think with the traffic is different because I think if we are sending traffic from 2 stations we want to make sure they are concurrent to test the AP performance with both stations working at the same time, right?
 

coxhaus

Part of the Furniture
Sorry I got stuck on TR-398 in general. Does not seem important to me.

Yes I would think you would want concurrent talking APs. Ping is controllable in the QOS hierarchical chain for only testing yes should work. You should not see lost packets.
 

thiggins

Mr. Easy
Staff member
There is no requirement to measure one-way delay in TR-398. The 6.5.1 Long Term Stability test has no requirement for time synchronization between AP and STA.

Remember that if MU-MIMO or OFDMA are not in use, only one device can transmit at a time on a channel.
 
Last edited:

coxhaus

Part of the Furniture
There is no requirement to measure one-way delay in TR-398. The 6.5.1 Long Term Stability test has no requirement for time synchronization between AP and STA.

Remember that if MU-MIMO or OFDMA are not in use, only one device can transmit at a time.
I thought that was why we use different channels so we can talk at the same time?
 

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top