Measurement of RTT in a LTE-EPC scenario

[RL]

Dear experts,

I would like to do the measurement of RTT in a LTE-EPC network with one UE and one eNB. Could you help me giving a reference in the tutorial or any link where I can study how I must do it?

Thanks very much in advance for your help.

[RL]

Hi!

I have been studying the tcp-variants-comparison.cc file but when I compile it has some problems. Anyway, my target is to get the rtt in a simple LTE-EPC network so I did in the attached file. My problem is the  medida.txt file is empty and it does not provide the value of rtt. I detected no access to the function RttTracer but I do not have any idea to find a solution. Could anyone help me in this issue?

Thanks very much.

[Konstantinos]

Hi,

You have to study the tutorial one more time to realise why just copying and pasting code from other scenarios without understanding what it does, most of the times will not work!

For example, you found a scenario that traces the RTT of a TCP socket. Ok, it looks promising and close to what you want to investigate. However, you left the traffic to UDP as the original example. Hence, no trace file. 

Fixing can be easy by creating TCP traffic. Other solution is to use PING application that generates RTT output (https://www.nsnam.org/docs/doxygen/classns3_1_1_v4_ping.html). 

For both of them there is a plethora of examples and discussions in the list.

Regards,

K.

[RL]

Hi Konstantinos,

Thanks very much for your answer. It is very appreciated for me. When I have a solution, I will explain it.

Regards,

RL.

[RL]

Hello again,

Finally, I obtained the RTT traces using  TCP traffic. The traced obtained has two columns, the first column has the instant of the simulation and the second one the value of RTT. I would like to verify if these values of RTT correspond to each packet sent but I do not find documentation to confirm this issue. Could anyone help me, please?

Thanks in advance.

RL.

[Konstantinos]

Hi,

(I assume you use the trace you have put in your previous post)

No, the trace file you got is not the RTT for each received packet. This is the estimated RTT and how it changes in the TCP.

For example, if for two consecutive packets the RTT does not change, then there will not be a trigger to print out that.

This is how traced-values work.

Regards,

K

[RL]

Hi Konstantinos,

Thanks very much for your quick and kind answer. It is very appreciated. Let me do an other question. I would like to measure the RTT of each packet, could you tell me if it is possible? If it is, could you recommend any links or example, please?

Thanks very much for your help.

Regards,

RL

[Konstantinos]

For TCP traffic, you can use the Tx/Rx trace sources to record the transmission time and reception time of each packet, and from that calculate the latency. That's one way traffic, i.e RTT/2. 

For UDP traffic, you can use the UdpClient/Server and analyse its logs. It can give you the latency of one way (again RTT/2).

Or finally, you can use Ping application that has a trace source for RTT.

Regards,

K.

[RL]

Hi Konstantinos,

Thanks very much again for your help, I am going to try to do your recommendations. I will return here to explain my solution.

Regards,

RL

[RL]

Hello,

Following the first posted simulation, my solution has been to use the class V4Ping for calculating the RTT but I have any doubts about the process. As far as I know, the latency can be measured by the time it takes for a small IP packet to travel from the terminal through the network to the internet server, and back. In my simulation I have performed several simulations but I don´t understand completely the result obtained.

It means:

Experiment 1: The address pinged is the remote host address (1.0.0.2) and the NodeContainer pinger added to node UE. Result: The RTT is not calculated.

Experiment 2: The address pinged is the remote host address (1.0.0.2) and the NodeContainer pinger added to node PGW. Result: The RTT is 20 ms.

Experiment 2: The address pinged is the remote host address (1.0.0.2) and the NodeContainer pinger added to remote host node. Result: The RTT is 0 ms. I do not understand it, because if I follow the definition of latency I should obtain the result of experiment 2 because the address pinged is the address of remote host.

Could anyone help me with some document or link for explaining it, please?

Thanks in advance for your collaboration.

Best regards.

[Konstantinos]

For experiments 2 and 3 the results are expected. For the first it may need more investigation why no packets are transferred. Perhaps if you attach the updated code we might be able to help.

The 20ms RTT on the second from PGW to 10.0.0.2 (remote host) confirm with the setup of the P2P link with 10ms delay.

 p2ph.SetChannelAttribute ("Delay", TimeValue (Seconds (0.010)));

The third is expected 0ms RTT because the node pings itself... 

Regards,

K. 

[RL]

 Hello Konstantinos,

First of all, thanks for your help, it is very important for me. With your previous explanation I understood completely the operation of ping. In my simulation (please you can find this attached), I know where the problem is for the first experiment, I can´t ping eNB node or UE node but the packets are received. It's like there was a border between EPC and LTE. I have been investigating my code and searching the reasons of this issue but I have not found any solution.

If you have any idea about the reason of this issue, please help me.

Thanks very much in advance.

Regards,

RL.

[RL]

Hello all,

   I have continued with my investigation about V4Ping. Is it possible V4Ping only works in a sub-network?

   I am working in the following LTE scenario:

   UE (7.0.0.2) ------- eNB (10.0.0.5) --------- PGW (1.0.0.1) ---------- RemoteHost (1.0.0.2)

   Do we have three sub-networks?
 
   My questions are in relation with the ping only works in sub-network 1.0.0.n

[Manuel Requena]

Hello Raul,

   I am working in the following LTE scenario:

   UE (7.0.0.2) ------- eNB (10.0.0.5) --------- PGW (1.0.0.1) ---------- RemoteHost (1.0.0.2)

   Do we have three sub-networks?

Not really. The LTE/EPC is more complex than just 3 subnets.

In [1], you can find how the LTE/EPC procotol stack are modeled in ns-3 and in [2], it is explained the data flow in downlink and uplink.

To summarize:

- The IP of the UE is used to communicate with the Internet, i.e. the subnet between the PGW and the Remote Host

- The Remote Host can reach the UE

- The IP of the eNB is used to established the S1 tunnels between the eNB and the SGW

- The UE cannot reach the IP subnet between the eNB and the SGW. They lived in different "worlds"

 

 
   My questions are in relation with the ping only works in sub-network 1.0.0.n

That is totally normal.

Best regards,

Manuel

[1] https://www.nsnam.org/docs/release/3.25/models/html/lte-design.html#epc-data-plane

[2] https://www.nsnam.org/docs/release/3.25/models/html/lte-design.html#s1

[Konstantinos]

Hi Manuel,

One observation with respect to your last comment. Perhaps you did not understand the initial question. The user asked if the ping only works on the 10.0.0.x subnet,which of course is incorrect. The remote host on 10.0.0.2 should be able to ping UEs in the 7.x.x.x subnet. If that's not happening then their is either an issue with the user scenario or with the LTE code. In either case it should be investigated starting with the user.

Regards
K.

[RL]

Hello Manuel and Konstantinos,

Thanks very much for your answers. Konstantinos understood my question, but I do not get my target and I continue in it. Few days ago I attached my code for receiving some help of somebody but it has not been possible so I do not know if the issue is in my scenario or in my LTE code.

Regards,

RL