Distributed Learning Based Joint Communication and Computation Strategy of IoT Devices in Smart Cities

Area of Science:

• Computer Science
• Electrical Engineering
• Artificial Intelligence

Background:

• Global urbanization drives the growth of Internet of Things (IoT) and smart cities.
• Mobile Edge Computing (MEC) offers low latency and high performance for IoT devices.
• Centralized resource allocation for numerous IoT devices in MEC leads to inefficient signaling and delays.

Purpose of the Study:

• To investigate the joint communication and computing policy for IoT devices in edge computing environments.
• To address the challenge of channel congestion and increased time overhead caused by simultaneous computation offloading.
• To develop an efficient, distributed algorithm for resource allocation in smart city IoT networks.

Main Methods:

• Utilized game theory to model the interactions between IoT devices and the MEC server.
• Proposed distributed Q-learning algorithms with two distinct learning policies.
• Conducted simulations to evaluate the performance of the proposed algorithms.

Main Results:

• The distributed Q-learning algorithms demonstrated rapid convergence.
• The proposed approach achieved a balanced solution for communication and computing resource allocation.
• Effectively mitigated channel congestion and reduced time overhead in simulated smart city scenarios.

Conclusions:

• Distributed Q-learning provides an efficient solution for resource management in edge computing for smart cities.
• The proposed joint communication and computing policy optimizes performance for a large number of IoT devices.
• This approach enhances the scalability and efficiency of IoT networks in urban environments.