NetSim Support Portal

Let us say the capacity of the network is U and let us say the average traffic generation rate is L. We say average since L need not be constant and can vary over time. To be precise, U represents the capacity of the bottleneck link between a particular source and a sink (destination) in a network. 

Assuming no other traffic in the network, the saturation condition is met when L = U.  The system is stable when L < U and unstable when L > U. 

Since the capacity of the network is generally not known prior to simulation, an important step toward analyzing a network is to find out U. One way to do this is to continuously increase the average traffic generation rate (a known quantity) till the condition L = U is hit. This can be inferred by observing the throughput. As the average traffic generation rate is increased the throughput will increase, till the saturation condition is reached. Once this threshold is crossed the throughput will remain flat as the average traffic generation rate is increased. This can be seen in the figure below.

Fig: Throughput vs. generation rate, where the network saturation is 10 Mbps

L can be increased by decreasing the inter-packet arrival time in the application configuration. We recommend you use CBR or Custom (with constant packet size and constant inter-packet arrival time distributions) application types.  As L approaches U packets start to get queued. Since each new packet has to wait longer than the current packet the per-packet delay will increase. 

Fig: Average queue as a function of generation rate. Packet inter-arrival times are exponentially distributed. 

It is a good practice to identify the saturation throughput for a network, before attempting more advanced performance analysis.

Configuring Full Buffer Traffic in NetSim based on saturation throughput

Full buffer traffic refers to a network traffic scenario where the network nodes always have packets waiting in their buffer queues, and there are always packets to be transmitted. In other words, the buffer queues are always full. Full buffer traffic can be modeled in NetSim by ensuring that the generation rate is higher than the capacity. 

Let us now understand how to model full buffer traffic in a 5g scenario in NetSim. 

1. We create a 5G scenario as shown in the figure.

2. We set the gNB properties as shown in the table

.

3. For all wired nodes we set link speed = 10 Gbps, Propagation delay =  0, and BER = 0.

4. We configure a downlink CBR application from the remote server to the UE. The generation rate is varied as shown in the table below. 

5. Run the simulation for 10 sec with different application generation rates.

Results

We observe that the saturation throughput of the network is 419.91 Mbps. Generating traffic above this value will ensure that the gNB buffer is always full. 

Configuring full buffer traffic using a "large file"

Some users prefer to use a "large file" to model full buffer traffic. How can this be done? From the above section, we know that the saturation throughput is 419.91 Mbps. This means in each second the network can handle 419.91/8 = 52.48 MB (megabyte) of traffic. Hence if we send data for 2 seconds it can transfer a file of size 104.96 MB. Thus for a 2-second data transfer time, if we set the file size > 104.96 MB, let's say 105 MB then this would also lead to a full buffer in the gNB.  We show this by creating the same scenario and modeling a large file transfer.

Note:  

• NetSim has two input parameters for configuring file transfer. They are (i) File size and (ii) inter-file arrival time. To limit the transfer to one large file we set the inter-file arrival time to be greater than the simulation time. 
• The FTP application is started at 1s since cell selection and attach procedures take up some time initially. We, therefore, run the simulation for 3s, allowing for a data transfer time of 2s i.e., 3s - 1s. 

Results

Useful links

1. 5GNR Documentation v13.3: 5G NR - NetSim Help Centre (tetcos.com)