Implementation and validation of random exponential marking (REM) in ns-3

The problem of bufferbloat has reignited interest in studying Active Queue Management (AQM) algorithms. Significant efforts have been taken by AQM and Packet Scheduling Working Group at IETF to bring more awareness about the performance benefits of deploying AQM algorithms in the Internet. However, experimental analysis of these algorithms is necessary prior to real time deployment. Network simulators like ns-3 are useful tools to perform such preliminary studies. Random Exponential Marking (REM) is one of the popular AQM algorithms. It decouples congestion measure from performance measure, and aims to stabilize the performance measure around the target queue length regardless of the number of users. This paper presents the implementation of a new model for REM in ns-3. The correctness of the proposed model has been validated by comparing the results obtained from it, to those obtained from the ns-2 model developed by authors of REM.