About the TDMA model packet drop rate

[jhonzy0928]

Why is it that when I use the default TDMA model to build the network topology, the connectivity between the nodes is particularly poor? When I use the ping command to check the packet loss rate between the nodes, I find that the packet loss rate between them can reach 20%-30%, which is exactly what we don't want to happen. I would like to ask how the packet loss rate of TDMA model can be reduced to below 5%.

[sgalgano]

Have you checked the statistic tables to see why you are dropping OTA frames? For the TDMA model, pay attention to RxSlotStatusTable and TxSlotStatusTable.

[jhonzy0928]

Thank you for your reply. What I want to say is that I agree with you and do notice it, Figure 1: TxSlotStatusTable has a very low percentage of frames lost, whereas RxSlotStatusTable has a very high percentage of frames lost. Looking at the other data entries again, Figure 2-3: There is no packet loss at all in the transport layer, but SlotErrorMissed is high in the MAClayer. The problem is that these days I try to create a network model with a very low packet loss rate using the TDMA model, but no matter how much I modify the configuration schedule and related parameters, the packet loss rate is still very high, even under the default TDMA model. To sum up, even in the example model (Demo-8) the packet loss rate can reach 30%, which is unacceptable for a reliable and real-time communication network, so what else can be done to reduce the packet loss rate?

[sgalgano]

Your system does not appear to be tuned correctly to run the experiment. Looking at your screenshot, it looks like you are running in a VM. Have you configured the virtualizer to make sure the VM has enough resources? Try running emane in LXC instances directly on your system.

[jhonzy0928]

I did demonstrate the sample model on a VM, and when I looked up the solution to this problem, I noticed that other developers had similar problems, and your answer confirms this. Next, I will allocate more resources according to the scheme you mentioned or directly run the example on the system. If it is successful, I will tell you the exciting news. Thank you again for your reply in your busy schedule.

[jhonzy0928]

I'm happy to tell you that when I run EMANE directly on my server, it works and the packet loss rate is very low. In the next stage I want to change the allocation of time slots under the TDMA model. For example, I want to dynamically allocate the size of time slots in the TDMA frame structure. But I don't know if EMANE supports this function or if you can give me some ideas.

[jhonzy0928]

It occurred to me that if I could dynamically detect traffic congestion in the network, such as detecting the cache of traffic queues, then I could dynamically publish TDMA events based on changes in network traffic. If this idea works, how can I implement it in EMANE?