An improved communication resource allocation strategy for wireless networks based on deep reinforcement learning

Abstract

Listen

Translate article

With the advent of 5G networks, user demand for high-speed, low-latency, and high-reliability services continues to grow. When traditional communication technologies cannot meet the needs, wireless network LoRa technology has emerged. Although LoRa has low power consumption, Long-distance, and other advantages, terminal nodes still face frequent data collection and energy consumption issues how to more efficiently combine the deep reinforcement learning method for LoRa wireless network communication and allocate resources reasonably and effectively. This paper proposes a communication channel resource allocation strategy based on deep reinforcement learning, the CL-LoRa strategy. It uses extended preamble and low-power interception technologies to achieve on-demand synchronization and low-power communication. The basic idea of this strategy is to detect channel quality based on CAD, coordinate node scheduling, and wireless channel allocation. The node will choose different ways to acquire the channel according to the current network load, namely CSMA-CA competition and dynamic duty cycle communication. In this way, the channel utilization rate is improved, and the energy consumption problem of the long-distance communication data volume is perfectly solved. The duty cycle access method is based on the imbalance of energy consumption in the Internet of Things. It uses the remaining energy of the remote central node to dynamically adjust the duty cycle of the node, wake up the working time of the node, and send more beacons to the sleeping node. Reduce the sleep delay of the node. Through theoretical analysis of CL-LoRa protocol performance, compared with DDC-LoRa protocol and ADC-LoRa protocol, CL-LoRa protocol can increase channel utilization by 9%, reduce terminal energy consumption by 1.6%, and increase throughput by 1.5%.