This project will develop a framework for controlled mobility in ad-hoc networks that integrates mobility and communication through the systematic combination of distributed cooperative control, resource allocation through reinforcement learning, and fault-tolerant network protocols. By exploiting mobility, large-scale, dynamic networks can be created that are rapidly reconfigurable in response to network traffic from various sources with different quality of service (QoS) requirements. This single framework unifies the design of data ferrying concepts for sparse or delay-tolerant networks where mobile nodes collect and carry data between other nodes in the system and the design of decentralized feedback controllers that optimize local real-time traffic flow in environments subject to interference. The result of the work proposed here will be increased flexibility in the deployment of ad-hoc networks on mobile platforms and improved network performance, in terms of capacity, throughput, delay, and other QoS measures, compared to systems that do not exploit node mobility.

Recent Research

Note! It is best to download the files and then play them.
• Globally maximize the min(SNRi) using fixed step size in the direction of the gradient
• Locally maximize the min(SNRi) using fixed step size in the direction of the gradient
• Globaly minimize the max(1/SNRi) using a fixed step size in the direction of the gradient
• Locally minimize the max(1/SNRi) using a scaled gradient
• A test variation showing smooth control using tau (chain delay) as performance function
• Locally equalize and maximize the neighbor links to maximize bi-directional communication with end node movement and node failure (note, actual chain capacity is miscalculated in this video due to a coding error)