Load Test Germany
-
Introduction:
As DT is reaching NB-IoT nationwide coverage in Germany and other NATCOs across its European footprint, often questions are asked about the actual performance of NB-IoT network.
How many device can attach to the network simultaneously? How much time do devices needs to attach to the network? what is the maximum data rate than can be achieved by a device? What are the limitations of network that will influence my use case? What is the maximum load capacity the network can handle?
Today we cannot give satisfying answers to above questions, but we will be able to answer all of these questions soon. We are aware about the importance of those questions and thus planned a Load test with the help of a specially designed load generator by SODAQ.
We will test the performance first in a closed lab environment to have better optimization possibilities and more control over various parameters. Later on the load test will be executed in the live network under real conditions.Load Generator:
Load generator is a box consisting of 10 NB-IoT communication modules, which are controlled by a master controller that collects results and send configuration to those 10 modules.
There are 20 boxes in total, allowing us to generate load from 200 communication modules simultaneously.
All boxes are configurable via dashboard, giving possibility to apply individual test sequence to individual boxes. Dashboard also reflects the measurement taken from each box in different form of graphs, making it easier for us to do the analysis of network Performance.
Test cases:
Several test cases have been planned to be executed. (e.g below)
• Attaching 10 devices first, then increasing the number to 20, 30, 40, 50….100, 200
• Starting data traffics of 10,20…100,200 bytes
All test cases will be executed in different combination to test the network capacity.Time plan for Load tests:
• Q3 Lab network testing
• End of Q3/Q4 Live network testingResults:
…… Loading soon! -
@Kashif-Shahab Nice! Very interesting. Question, did you consider using a UE simulator for this? For example this one: https://www.amarisoft.com/products/test-measurements/amari-ue-simbox/ It can emulate 1000’s of UE’s connecting at the same time.
-
@Stefan-de-Lange Hi Stefan, we wanted to use replicate real Scenario as much as possible, and wanted to use the chipset in devices that our Clients are using therefore for this time we didnt consider the Simulators.
We have load figures already from our Vendors, who did the testing using such Simulators. -
This post is deleted! -
This post is deleted! -
This post is deleted! -
Hello @Kashif-Shahab ,
This is really interesting.
What were the results of this test? Can you share more details on the outcomes?Thanks,
Regards,
Ahmed Elhaddad -
Hello @Ahmed-Elhaddad,
We do these tests in parallel to normal network maintenance and improvement measures and we can share some preliminary results in CE-0 conditions from measurements in our Labs.
• Up to 200 devices can attach successfully simultaneously within 10 minutes under the condition that they do not immediately send data after a successful attach.
• Average round trip time (RTT) increases with increasing number of parallel transmitting devices. A randomization of transmission attempts leads to reduction of RTT and potentially improvement of battery lifetime
• With regards to the initial attach, only a small number of devices should be activated simultaneously (max. 60 devices)
• With regards to the regular data transmission with good signal level the following was observed:
• Failure rates increase with more than 160 devices trying to send 100 byte of data within one second
• Introducing a randomization timer of at least 40 seconds solves this limitation
• Avoiding synchronized data transmission results in better share of coverage levels across all devices
• Implication - Customers should avoid synchronized attaches and data-transmission of their devicesBest,
Dominik