OPTIMIZING DRONE SWARM-BASED WIRELESS NETWORK USING STOCHASTIC MODEL AND CONTROLLER APPROACH
Izzat Al-Darraji,
Ali T.Hasan,
Salah sabeeh Abed Al Kareem
Abstract: Wireless control drone swarm systems are more and more widely preferred and be chosen for its potential to reduce the cost of communication network. Yet, the performance of such systems is largely compromised due to network inherent limitations, such as time-varying delays, packet dropouts and bandwidth constraints. The traditional control and scheduling schemes such as time-triggered communication and static bandwidth allocation are not able to provide an adaptive control under dynamic network conditions which results in inefficient control performances. This article discusses a distributed control architecture for autonomous drone swarms in presence of a Wireless Networked Control Systems (WNCS), where the wireless device underpinning some disruptive behaviors such as stochastic delays, packet loss and bandwidth availability. To this end, a joint optimization algorithm is developed, where stochastic model predictive control, event-triggered communication, and network-aware scheduling are integrated into a control framework. There is a decentralized local control for each agent using formation consensus-based control rules for tracking. The approach is adaptive to communication noise with quantized value estimation and decentralized triggering. Numerical simulations of the developed approach are developed in MATLAB which span over three case studies: (1) baseline Linear Quadratic Regulator (LQR) control in presence of packet loss and delay, (2) online model predictive control in the presence of uncertain dynamics and adaptive scheduling and (3) formation flying with swarming drones with consensus. The experiments show that the proposed method is robust, scalable, and bandwidth efficient. Average tracking and state errors are limited and communication loads are suppressed by event-triggered updates. Furthermore, robustness is demonstrated against the challenging operating environment by visualizing packet loss.
Keywords: Wireless Networked Control Systems, Drone Swarms, Stochastic Optimal Control, Event-Triggered Communication, Multi-Agent Coordination, Network Scheduling and Control Co-Design
DOI: 10.24874/PES08.02B.010
Recieved: Revised: Accepted:
UDC:
Reads: 0 
