WNOS-G2: A Learning-based Wireless Network Operating System With Autonomous Distributed Network Control

Abstract

A core problem in tactical wireless networks is to translate commander s intent, often expressed in terms of tasks that require timely and guaranteed response, into a set of policies - and eventually a set of distributed control actions - that can provide high throughput and reliable information gathering and delivery, sometimes in the presence of uncertainty and adversarial actions. This is often difficult because of the distributed nature of the network control problem and of the challenging battlefield conditions. First, tactical wireless networks may consist of thousands of devices that need to be networked in an ad hoc manner with only limited or no infrastructure at all. Second, the radio spectrum can be very dynamic and severely congested because of the presence of movements of warfighters, adversarial jamming as well as changes in the battlefield landscape. Third, existing wireless networks are inherently hardware-based and rely on closed and inflexible architectures. Such inflexible hardware-based architectures impose significant challenges into adopting new wireless networking technologies, and prevent the provision of a true networking-as-a-service vision able to adapt to commander s intents. Our vision is simple. These challenges cannot be addressed without a new approach to design and manage tactical wireless networks that will result in distributed, optimal, and autonomous control with guaranteed delivery in the presence of adversarial attacks. The objective of this proposal is to establish an applied research program to develop the core building blocks for a new autonomous network management system - rooted in learning theory and non-linear optimization - for tactical wireless networks, based on which the commander s intent can be translated, in an automated manner, into distributed optimal network operating policies. This new framework will build (and significantly improve upon) a new principled software-defined networking framework, the Wireless Network Operating System (WNOS), that has been developed in our previous ONR-funded research over the past three years.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 17, 2020
Source ID
N000142012132

Entities

People

  • Tommaso Melodia

Organizations

  • Northeastern University
  • Office of Naval Research
  • United States Navy

Tags

Fields of Study

  • Computer science

Readers

  • Computer Networking
  • Joint Military Operations and Doctrine.
  • Neural Network Machine Learning.