Optimize Communication Architecture in Dynamic Combat Environment via Online Learning
Hao Yuan, Тао Чен, Bangbang Ren, Mengmeng Zhang, Luo Xue-shan · IEEE Systems Journal · 2025
The application of artificial intelligence, Big Data, and other advanced technologies has dramatically improved the intelligence level of the combat system-of-systems and accelerated the combat rhythm, which requires higher decision speed in the support of high-quality combat communication architecture. In reality, due to the poor infrastructure conditions on the battlefield, the communication services of the combat units are usually provided by the communication units with limited communication resources. Thus, figuring out an efficient method to share the scarce communication resources among massive combat units becomes crucial. However, it is challenging to efficiently construct the connection relationship and allocate communication resources to the operational units because of the differences in communication requirements and the randomness of location movement of combat units, i.e., unable to obtain battlefield environmental information in advance. In this article, we propose an online learning (OL)-based combat communication architecture construction method, which can estimate the current state of the battlefield environment by interacting with it and dynamically construct connection relationships and allocating communication resources according to the needs and locations of operational units, so as to maximize the QoE. The evaluation results demonstrate that our proposed OL-based approach is capable of constructing the combat communication architecture in a flexible and efficient manner, surpassing existing methods in terms of efficiency and fairness by significantly enhancing the total QoE up to twice as much compared to baseline methods.