5G toolbox IP packet latency calculation

How to calculate the application layer latency for each UE in the simulation?

I am using NR FDD Scheduling Performance Evaluation Example. It should be defined as the time when the packet arrives to UE (in the buffer) minus the time when the packet arrives to BS for UL. Do you have some hints on how I can track those times?

Answer 1

[Ayush]

Hey Nadezda,

I understand that you are using NR FDD scheduling performance evaluation example and wants to calculate application layer latency for each UE in the simulation for which you need the time at which UE packet arrives and time at which packet arrives at BS for UL. You are looking for a way to track these times.

Based on the information provided in the documentation for the MATLAB NR FDD Scheduling Performance Evaluation Example, you can use the RLC logs and RLC statistics to track the packet arrival times and calculate the application layer latency. Here's how you can do it:

1. Packet Arrival at UE: From the RLC statistics for each UE, you can use the "ReceivedPackets" or "ReceivedBytes" field to determine when a packet arrives at the UE from the MAC layer. Each row in the UE RLC statistics table represents a logical channel of a UE. You can track the timestamp associated with the row where "ReceivedPackets" or "ReceivedBytes" is greater than zero.

2. Packet Arrival at BS: Similarly, from the RLC statistics for the gNB, you can use the "TransmittedPackets" or "TransmittedBytes" field to determine when a packet is transmitted from the RLC layer to the MAC layer at the BS. Each row in the gNB RLC statistics table represents a logical channel of a UE. You can track the timestamp associated with the row where "TransmittedPackets" or "TransmittedBytes" is greater than zero.

Once you have the packet arrival times at both the UE and the BS, you can calculate the application layer latency for each UE using the formula:

ApplicationLayerLatency = UEArrivalTime - BSArrivalTime

Note: Make sure that the timestamps are in the same time units (e.g., milliseconds) before performing the subtraction.

For more information, you can refer to the documentation: NR FDD Scheduling Performance Evaluation - MATLAB & Simulink - MathWorks India

Hope this helps!

[Nadezda]

Thank you for you answer! Indeed, part of the application layer latency can be taken from RLC logger. However, the ApplicationLayerLatency = UEArrivalTime - BSArrivalTime gives only transmission part (and usually is 1 ms) while application layer latency should also include time when packet arrives at the UE/BS buffer depending on uplink/downlink transmission.

For full application layer latency, I track the time when the packet arrives at MAC layer of e.g., UE buffer (the easiest way to track it) and stop tracking the packet when it is receieved by BS at RLC layer (as you suggested). The time between both of them is the closest aplication layer latency I could implement using this example code.