Delay Minimization in UAV-Enabled MEC Networks Using Q-Learning

Mobile edge computing (MEC) is a promising solution for the ever-increasing computational demands of the ubiquitous Internet of things (IoT) applications. However, efficient resource management algorithms are needed to satisfy the quality-of-service (QoS) requirements of delay-sensitive IoT systems. In this paper, we propose a new Q-learning-based load-balancing resource management (QLLB) algorithm to reduce the computational delay in a MEC system. Unlike existing algorithms, the proposed algorithm optimally assigns tasks to unmanned aerial vehicles (UAVs) considering load balancing, and employs the Q-learning algorithm to achieve efficient CPU usage management at the MEC server level. Results show that the QLLB algorithm exhibits a considerable improvement in the overall performance compared to the baseline and round-robin algorithms. Specifically, under heavy load conditions, we observe a 68% reduction in the total delay and a 75% reduction in the overall algorithm execution time with respect to the baseline algorithm. Furthermore, we observe a 59% improvement in the combined UAV selection and task data transmission delay.