Home Mamoutou Diarra

Mamoutou Diarra

Mamoutou Diarra

I am a P.h.D. student in the DIANA team at INRIA under the supervision of Thierry Turletti and Walid Dabbous. My P.h.D is funded by Ekinops via the CIFRE program.

I received a Post Master’s degree in Network Security from Eurecom in 2018 (Sophia-Antipolis, France) as well as an Engineering degree in Telecommunication from ESPRIT in 2016 (Ariana, Tunisia ). Before starting P.h.D studies, I was working as an R&D engineer (in NFV/SDN and Transport-layer optimization) at Ekinops. On behalf of Ekinops, I also contributed to the 5GENESIS and SAT5G European Union projects, particularly on topics involving Multi-path TCP (MPTCP), Multi-path QUIC (MPTCP), TCP acceleration and bakhaul traffic splitting (between Satellite and terrestrial links).

In the DIANA team, my research mainly focuses on enhancing TCP/QUIC transport-layer mechanisms (e.g., Loss Detection, Flow Control, Steady state behavior, etc.) and addressing common TCP issues in cellular networks (e.g., bufferbloat, radio bandwidth under-utilization, etc.) by combining specific Multi-Access Edge Computing (MEC) capabilities with relevant transport-layer metrics. With this approach we can easily exploit RAN information at the transport-layer without requiring any modifications in the end-user’s device or in the 4G/5G protocol stack (as opposed to existing cross-layer solutions).

Research Interests

  • TCP/QUIC Optimization in Cellular Networks
  • QoS mechanisms in Cellular Networks improvement (RAN & CN)
  • Congestion Control Algorithms
  • Cross-layer L4 Optimization techniques
  • Multi-access Edge Computing (MEC)

Projects

  • MELD: (MEC-based Edge Loss Discrimination) is a novel server-side loss discrimination mechanism that leverages the Radio Network Information Service (RNIS) offered by MEC in order to discriminate packet losses based on real-time RAN statistics. the current version of MELD is based on picoquic, OpenAirInterface and FlexRAN.
  • RAPID: (RAN-Aware Proxy-based flow control for HIgh througput and low Delay) is a TCP Performance Enhancing Proxy (PEP) that mitigates self-inflicted bufferbloat and maximizes radio link utilization in 4G/5G cellular networks, by intelligently overriding the Receive Windows of TCP flows (in the uplink direction). Basically, it intercepts TCP flows and distributes proportionally the available RAN bandwidth among them based on their respective demands. It relies on the RNIS API in order to estimate the aggregated RAN bandwidth and on packets arrival rates in order to capture the demand of each intercepted flow. Currently RAPID is implemented on Ubuntu as a Kernel module. This implementation is based on pepsal, netfilter, FlexRan and OpenAirInterface
  • SIGMA: (Simple Increase in Goodput based on MEC Awareness) is a simplistic uplink-oriented Congestion Control Algorithm (CCA) that takes advantage of the known location of the bottleneck (which is the RAN in case of MEC and most CDN scenarios) and the availability of radio information at the UE in order to enhance the uplink TCP performance (in terms of throughput and delay). Our current evaluations in NS3 show that SIGMA outperforms BBR, Cubic and NewReno both in terms of throughput and delay. SIGMA is currently implemented in NS3 but we are considering a future real-world implementation in order to validate its performance.

Publications

  • Mamoutou Diarra, Walid Dabbous, Amine Ismail and Thierry Turletti, “Cross-layer Loss Discrimination Algorithms for MEC in 4G networks”, 2021 IEEE 22nd International Conference on High Performance Switching and Routing (HPSR), 2021, pp. 1-6, doi: 10.1109/HPSR52026.2021.9481843.
  • Mamoutou Diarra, Walid Dabbous, Amine Ismail, and Thierry Turletti, “RAN-aware Proxy-based Flow Control for High Throughput and Low Delay eMBB”, In Proceedings of the 24th International ACM Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWiM ’21), 2021, pp. 41–50, doi: 10.1145/3479239.3485717.
  • Mamoutou Diarra, Walid Dabbous, Amine Ismail, Brice Tetu, and Thierry Turletti, “Rapid: A Ran-Aware Performance Enhancing Proxy for High Throughput Low Delay Flows in Mec Networks”, Accepted and in the process of being published in the Computer Networks journal (currently available at http://dx.doi.org/10.2139/ssrn.4082865.

Contact

  • Organizations: INRIA and Ekinops
  • Research Center: Sophia Antipolis – Méditerranée
  • Project teams: DIANA (at Inria) ; Study and Research (at Ekinops)
  • E-mail (Inria): mamoutou.diarra@inria.fr
  • E-mail (Ekinops): mamoutou.diarra@ekinops.com
  • Tel: +33 4 93 00 1662
  • Address: INRIA, 2004 route des Lucioles, B.P. 93, 06902 Sophia Antipolis, France

Comments are closed.