Method of increasing the efficiency of a local radio network based artificial intelligence

Abstract

Relevance. Ensuring the survivability of local radio networks (LRN) in conditions of electronic warfare is a critical scientific and technical task. Classical frequency hopping (FH) loses its effectiveness against intelligent jammers, which forces us to look for proactive solutions based on artificial intelligence. Object of research. The process of functioning, protection and dynamic allocation of frequency resources in multi-node local radio networks under conditions of intelligent interference. Purpose of the article. Development of a dynamic FH method based on reinforcement learning (RL) for autonomous counteraction to adaptive jamming, maximizing throughput and minimizing delays and collisions. Research results. A mathematical formulation of the problem is proposed in the form of a partially observable Markov process (POMDP) with strict constraints on power, delays, and intra-network collision probability. A multi-component instantaneous reward function is developed for balanced optimization of throughput and energy efficiency. A computer simulation in Python is performed, which compares the RL agent with the basic random reconfiguration method. It is experimentally proven that if the basic method demonstrates critical drops in the probability of successful transmission to 0.70 due to vulnerability to adaptive attacks, then the RL method quickly identifies jammer patterns and stabilizes the success rate at a level above 0.90, significantly reducing the number of retransmissions and energy consumption. Conclusions. Integration of RL algorithms directly into the physical and link layers allows you to move from static to fully autonomous protection. The proposed method effectively neutralizes adaptive threats, providing continuous communication in real time and in complex conditions without prior knowledge of the enemy's algorithms. This is a key factor for increasing the reliability of modern wireless systems, industrial Internet of Things and critical infrastructure in conditions of high uncertainty of the electronic environment.

Keywords: local radio network, reinforcement learning, intelligent jamming, countermeasures, Markov decision-making process, artificial intelligence, electronic warfare, dynamic frequency allocation.