BBRv1 does not perform well in variation of bandwidth scenario compared to cubic, so we have tried to improve the BBRv1 in this scenario. The amelioration in TCP is based in the packet- processing delay, as we know that bbr use round-trip time to calculate the BDP multiplied by the bottleneck bandwidth, but we have a little time that we can profit in sending of data, this time is known as the packet-processing time, when the sender receives the acknowledgments of the transmitted data, it waits a time to process the receive data and to process the next data that it want to send, after that the data is transmitted, so the link is not perfectly utilized in this processing time, we have suggested to adds this time in the calculation of the minimum round-trip time or as it's called the RTprop to get more improvement in throughput.

#Instructions:

How to Run:

Just compile and install is as a module. Change the default Congestion Algorithm to "NR_BBRv1"

Above development was based on the MPTCP v0.95.0 deployed in Linux kernel version of v4.19.206

Regards,

Computer Systems Enginner Drici Oussama.

For any information contact me at o.drici@esi-sba.dz