Intelligent Automation, Inc. — Department of Defense SBIR Phase II: A18-051

Description

Tactical networks face congested and contested environments characterized by intermittent connectivity due to severe jamming and interference conditions. In addition, tactical radios are limited by different levels of resources such as energy, computation, and communication. Effective coordination and cooperation between agents (namely, network nodes) is necessary to manage information flows and satisfy the requirements of tactical communications. As tactical networks need to cover large areas (e.g., brigade size), they need self-organizing, self-healing and self-optimization capabilities such that nodes join, leave, and rejoin the network without manual intervention, links are repaired and rerouted autonomously, network and radio resources are utilized efficiently, and anti-jamming modes, are realized in a timely manner. IAI will address these challenges with the design and implementation of Coordination and Cooperation in Operational Networks (COCOON). COCOON enables nodes to make intelligent transmission decisions based on novel reinforcement learning (Q-learning) and deep reinforcement learning algorithms (deep Q-learning) that utilize network and radio resources, including the use of cooperative (friendly) jammers in the congested and contested environments. The outcome of COCOON is a set of network layer protocols that enable robust operation in a contested and congested environment, and an SDR implementation.