Nowadays, mutual interference among automotive radars has become a problem of wide concern. In this paper, a decentralized spectrum allocation approach is presented to avoid mutual interference among automotive radars. Although decentralized spectrum allocation has been extensively studied in cognitive radio sensor networks, two challenges are observed for automotive sensors using radar. First, the allocation approach should be dynamic as all radars are mounted on moving vehicles. Second, each radar does not communicate with the others so it has quite limited information. A machine learning technique, reinforcement learning, is utilized because it can learn a decision making policy in an unknown dynamic environment. As a single radar observation is incomplete, a long short-term memory recurrent network is used to aggregate radar observations through time so that each radar can learn to choose a frequency subband by combining both the present and past observations. Simulation experiments are conducted to compare the proposed approach with other common spectrum allocation methods such as the random and myopic policy, indicating that our approach outperforms the others.

中文翻译：

---

使用强化学习避免相互干扰的分散式汽车雷达频谱分配

如今，汽车雷达之间的相互干扰已成为人们广泛关注的问题。在本文中，提出了一种分散的频谱分配方法，以避免汽车雷达之间的相互干扰。尽管在认知无线电传感器网络中广泛研究了分散式频谱分配，但观察到使用雷达的汽车传感器面临两个挑战。首先，分配方法应该是动态的，因为所有雷达都安装在移动的车辆上。其次，每个雷达不与其他雷达通信，因此它的信息非常有限。使用机器学习技术，即强化学习，是因为它可以在未知的动态环境中学习决策制定策略。由于单次雷达观测不完整，一个长短期记忆循环网络用于通过时间聚合雷达观测，以便每个雷达可以通过结合当前和过去的观测来学习选择频率子带。进行仿真实验以将所提出的方法与其他常见的频谱分配方法（例如随机和近视策略）进行比较，表明我们的方法优于其他方法。