Home DrIVE Associated Team: Distributed Intelligent Vehicular Environment – Enabling ITS through programmable networks

DrIVE Associated Team: Distributed Intelligent Vehicular Environment – Enabling ITS through programmable networks

DIANA_Capture     

Objectives

Transportation systems are part of our society’s critical infrastructure and are expected to experience transformative changes as the Internet revolution unfolds. The automotive industry is a notable example: it has been undergoing disruptive transformations as vehicles transition from traditional unassisted driving to fully automated driving, and eventually to the self-driving model. Communication technology advancements such as 5G and support for vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) communication have been one of the key enablers of next-generation transportation services, also known as Intelligent Transport Systems (ITS). However, ITS services and applications pose significant challenges to the underlying communication and network infrastructure due to their stringent low latency, reliability, scalability, and geographic decentralization requirements. The DrIVE associated team proposal aims at addressing such challenges by:

  • developing a programmable network control plane that will dynamically adjust to current environment conditions and network characteristics to support ITS’ scalability, quality of service (QoS), and decentralization requirements, and
  • applying the proposed distributed network control plane framework to ITS services and applications, such as road hazard warning, autonomous- and self-driving vehicles, and passenger-centric services (e.g., infotainment and video streaming).

Main contributions

  • Network planning and provisioning:  We proposed a framework that optimizes the deployment cost of 5G networks. It is based on a Mixed Integer Quadratically Constrained Programming (MIQCP) model that optimizes the deployment cost of 5G network functions while performing adequate provisioning to address user demand and performance requirements. We use two realistic scenarios to showcase that our framework can be applied to different types of deployments. This work has been presented at the IEEE ICC conference in 2018 [C1].
  • Message dissemination in ITS: We built upon our prior work D2-ITS [rel-0] and extended it with handover and load balancing capabilities. More specifically, the new handover feature allows a controller to automatically “delegate” control of a vehicle to another controller as the vehicle moves. We showcased handover and load-balancing features using the Mininet-Wifi network simulator/emulator and presented this contribution at the ICCCN conference in 2018 [C2].
  • Dynamically Distributed Network Control for Intelligent Transportation Systems: We have been developing a dynamically distributed and decentralized network control plane framework to support autonomous or semi-autonomous driving. The proposed framework is based on a hierarchy of controllers in which local controllers are placed closer to vehicles in order to better satisfy the different quality-of-service requirements of cooperative driving services (e.g., ultra-low latencies) as well as adapt to dynamic network- and environment conditions. We presented the proposed control plane architecture and discussed our current implementation, as well as extensive experiments using realistic scenarios in [R6].
  • Dynamic Controller Assignment in Software Defined Internet of Vehicles through Multi-Agent Deep Reinforcement Learning: We proposed a novel dynamic controller assignment algorithm targeting connected vehicle services and applications, also known as Internet of Vehicles (IoV). Our approach considers a hierarchically distributed control plane, decoupled from the data plane, and uses vehicle location and control traffic load to perform controller assignment dynamically. We model the dynamic controller assignment problem as a multi-agent Markov game and solve it with cooperative multi-agent deep reinforcement learning. Simulation results using real-world vehicle mobility traces show that the proposed approach outperforms existing ones by reducing control delay as well as packet loss. This work has been published to the IEEE TNSM journal [J1].
  • Harnessing UAVs for Fair 5G Bandwidth Allocation in Vehicular Communication via Deep Reinforcement Learning: Terrestrial wireless infrastructure-based networks do not always guarantee that their resources will be shared uniformly by nodes in vehicular networks mostly due to the uneven and dynamic distribution of vehicles in the network as well as path loss effects. We leveraged multiple 5G unmanned aerial vehicles (UAVs) to enhance network resource allocation among vehicles by positioning UAVs on-demand as flying communication infrastructure. We proposed a deep reinforcement learning (DRL) approach to determine the position of UAVs in order to improve the fairness and efficiency of network resource allocation while considering the UAVs’ flying range, communication range, and limited energy resources. In particular, we used a parametric fairness function for resource allocation that can be tuned to reach different allocation objectives ranging from maximizing the total throughput of vehicles, maximizing minimum throughput, as well as achieving proportional band-width allocation. Simulation results show that the proposed DRL approach to UAV positioning can help improve network resource allocation according to the targeted fairness objective. This work has been published to the Transactions on emerging telecommunications technologies journal [J3].
  •  Machine Learning for Next-Generation Intelligent Transportation Systems:  Intelligent Transportation Systems, or ITS for short, includes a variety of services and applications such as road traffic management, traveler information systems, public transit system management, and autonomous vehicles, to name a few. It is expected that ITS will be an integral part of urban planning and future cities as it will contribute to improved road and traffic safety, transportation and transit efficiency, as well as to increased energy efficiency and reduced environmental pollution. On the other hand, ITS poses a variety of challenges due to its scalability and diverse quality-of-service needs, as well as the massive amounts of data it will generate. We explored the use of Machine Learning (ML), which has recently gained significant traction, to enable ITS. We wrote a comprehensive survey of the current state-of-the-art of how ML technology has been applied to a broad range of ITS applications and services, such as cooperative driving and road hazard warning, and identified future directions for how ITS can use and benefit from ML technology. This survey has been submitted to the IEEE TNSM journal [J2].

Testbeds and Software

We have also been developing a variety of platforms including  network testbed, emulator, and simulator to make it possible to evaluate ITS communication mechanisms developed within DrIVE in realistic, yet reproducible scenarios:

  • D2-ITS simulation code used for [R3]: Git link.
  • R2lab testbed: we have enhanced the performance of the experiment control tool nepi-ng that it used to easily create and perform wireless scenarios within the testbed. The tool has been demonstrated at the ACM Mobicom Wintech 2018 workshop and a paper has been presented in the same place too [rel-2,rel-3]. Then, we have proposed a way to mix experimentation and simulation in the same scenario to enable for instance real wireless transmissions between R2lab nodes and mobile simulated nodes running on the ns-3 simulator [rel-1].
  • Mininet-Wifi network emulator: we have implemented two main extensions to the Mininet-Wifi network emulator – (1) Wireless nodes or vehicles with access point association can report their connectivity status to the SDN controller in terms of signal strength and other parameters. This new feature will be used for handover of vehicles which can be initiated by an SDN controller or by the vehicle itself; (2) Inter-controller communication through a publish-subscribe interface to allow state exchange between SDN controllers. Using publish-subscribe communication will facilitate different communication models, e.g., one-to-one, one-to-many, many-to-one, many-to-many. In order to add the new Mininet-Wifi functionality described above, we integrated tools and technologies such as Scapy and WPA Supplicant. Scapy is used for transmission and manipulation of 802.11 beacons. WPA Supplicant allows control of roaming nodes and their association in a wireless network based on IEEE 802.11.

Joint Publications

Related Publications

  • [rel-5] Houssam Elbouanani, Chadi Barakat, Guillaume Urvoy-Keller, Dino Lopez-Pacheco, “Collaborative Traffic Measurement in Virtualized Data Center Networks“, Poster at IEEE CloudNet’19, 4-6 Nov., Coimbra, Portugal.
  • [rel-4] K. Matsuzono, H. Asaeda, Indukala Naladala, Thierry Turletti, “Efficient Pull-based Mobile Video Streaming leveraging In-Network Functions“, IEEE 17th Annual Consumer Communications & Networking Conference (CCNC), Jan 2020, Las Vegas, USA.
  • [rel-3] Thierry Parmentelat, Thierry Turletti, Walid Dabbous, Naoufal Mahfoudi, Francesco Bronzino, “nepi-ng: an efficient experiment control tool in R2lab“, ACM WiNTECH 2018 – 12th ACM International Workshop on Wireless Network Testbeds, Experimental evaluation & CHaracterization, Nov 2018, New Delhi, India. pp.1-8.
  • [rel-2] Yonathan Bleyfuesz, Thierry Parmentelat, Thierry Turletti, Farzaneh Pakzad, Mohamed Mahfoudi, Walid Dabbous, “Demo: Using nepi-ng for Mesh Networks Experiments“, The 12th ACM International Workshop on Wireless Network Testbeds, Experimental evaluation & CHaracterization (WINTECH), Nov 2018, New Delhi, India.
  • [rel-1] Indukala Naladala, “Integrating R2lab with ns-3” B.T. in CS and Engineering Internship report, NITK Mangalore, July 2018.
  • [rel-0] Anuj Kaul, Katia Obraczka, Mateus Santos, Christian Rothenberg, Thierry Turletti, “Dynamically Distributed Network Control for Message Dissemination in ITS“, IEEE/ACM DS-RT 21st International Symposium on Distributed Simulation and Real Time Applications, Oct 2017, Rome, Italy (prior work to DrIVE associated team).

Related Thesis/Reports

  • Houssam Elbouanani (M.Sc. Ubinet UCA 2019), M.Sc. thesis “Measurement as a Service in Modern Data Centers“, August 2019.
  • Mohamed Naoufal Mahfoudi, Thierry Parmentelat, Thierry Turletti, Walid Dabbous, Towards Reproducible Wireless Experiments Using R2lab, Research Report, June 2019.
  • Y. Bleyfuesz,  (Master 1 Informatique UNSA 2018) on Experimental Comparison between OLSR and BATMAN protocols using R2lab.
  • Indukala Naladala (B.T. in CS and Engineering, NITK Mangalore 2018) on Integrating R2lab with ns-3.

Visits between Partners

  • July 17-31, 2018: Katia Obraczka 2-weeks visit at Inria.
  • December 3-7, 2018: meeting at UNICAMP (visit of Katia Obraczka and Thierry Turletti)
  • July 10-12, 2019: meeting at Inria (visit of Christian Esteve Rothenberg and Katia Obraczka)
  • Oct 21-24, 2019: meeting at UNICAMP (visit of Tingting Yuan and Thierry Turletti)
  • Oct 21 – Nov 8, 2019: Tingting Yuan 3-weeks visit at UNICAMP
  • All visits scheduled on 2020 cancelled due to COVID-19 containments.

Participants

DIANA at Inria Sophia Antipolis, FRANCE:

  • Thierry Turletti (PI): Senior Researcher. Network architectures, Wireless and Testbeds skills.
  • Chadi Barakat: Senior Researcher. Measurement, QoE and network modeling skills.
  • Thierry Parmentelat: Senior Software Engineer. Software and Testbed skills.
  • Osama Arouk (up to September 2018): Postdoc Labex@UCN on 5G NFV slicing.
  • Tingting Yuan (from Nov 2018): Postdoc Inria on Software-Defined VANETs.
  • Naoufal Mahfoudi (up to Oct 2019): PhD student, with expertise on experimentation in the R2lab wireless testbed.
  • Houssam Elbouanani (from Dec 2019): PhD student, with network monitoring skills.
  • Houssam Elbouanani (Mar – Aug 2019): M.Sc. student from UCA Ubinet on traffic measurement, see [rel-5].
  • Indukala Naladala: (June-July 2018) B.T. student from NITK Mangalore, 2-months internship on integrating ns-3 to R2lab, see [rel-4].
  • Adeel Khalid Siddiqui: (Nov – Dec 2019) M.Sc student from Ubinet UCA, France, PFE on integrating mininet-wifi within R2lab.

INTRIG at UNICAMP, Campinas-SP, BRAZIL :

  • Christian Esteve Rothenberg (PI): Assistant Professor, head of  INTRIG team.  Wireless, Security and Network architectures skills.
  • Ramon Dos Reis Fontes: Postdoc, Wireless SDN and mininet-wifi skills.
  • Gyanesh Kumar Patra: PhD student, control software design for networking and SDN skills.
  • Wilson Borba da Rocha Neto: PhD student, SDN and ITS skills.
  • Daniel Harrison: MSc student, ITS skills.

Ericsson Research, Indaiatuba-SP, BRAZIL (Satellite Partner):

  • Mateus Augusto Silva Santos (PI): Experienced Researcher. Ad hoc networks and SDN skills.
  • Pedro Henrique Gomes: Experienced Researcher. Industrial IoT skills.
  • Daniel Harrison: Researcher at Ericsson and regular MSc student at Unicamp since Aug/2017, expected to defend in the 2nd semester of 2019.

UC Santa Cruz, CA, USA (Other Partner):

  • Katia Obraczka (PI): Professor, head of i-NRG team, Network architectures skills.
  • Anuj Kaul: PhD student, SDN and ITS skills.
  • Lie Xue: PhD student, Security and SDN skills.

Comments are closed.