Data Integrity for a Reinforcement Learning-Based Indoor Drone Control in High Network Traffic

Abstract

During the high demand of data, computerized networks consistently face challenges to fluently provide real- time responses. This research defines operational factors to investigate the integrity and reliability of reinforcement learning (RL) for controlling indoor drones under network conditions of high traffic. Through this research, the proposed framework integrates RL algorithms with advanced 3D telemetry and secure privilege escalation to maintain efficient flight control and communication integrity in complex indoor areas. Using MATLAB and Simulink for a user friendly simulations and real- time testing, the system dynamically adapts flight paths and enhances communication policies to solve the packet loss and transmission latency issues. Experimental results demonstrate telemetry update rates of approximately 25Hz, a quasi- centimeter far positional accuracy, and communication packet error rates below 1%, validating the effectiveness of the approach. Comparative analysis between a custom data streaming method and the MAVLink protocol confirms enhanced telemetry integrity and error detection under heavy network load. The architecture also reduces operator cognitive load by providing intuitive 3D visualization and secure access controls. Thus, this work establishes a scalable and secure RL- based framework for reliable indoor drone operation in high- traffic network scenarios.