Double‐deep Q‐learning‐based handover management in mmWave heterogeneous networks with dual connectivity

Abstract

Millimeter wave (mmWave) technology, with its abundant spectrum resources and ultra‐high bandwidth, plays a crucial role in meeting the ultra‐high throughput requirements of future 6G communication and supporting a large number of device connections. However, mmWave signals are highly susceptible to congestion, and the channel quality in mmWave links can be extremely unstable, posing a significant challenge to mobility management in mmWave networks. In this article, we present a deep reinforcement learning‐based dual connectivity (DC) handover strategy for mmWave cellular networks. The DC architecture ensures a continuous connection to the core network, reducing the likelihood of frequent service interruptions. Additionally, the proposed handover strategy employs deep reinforcement learning to select the optimal handover target node, further enhancing system performance. Compared to traditional single connectivity (SC) handover strategies, DC involves more intricate handover trigger conditions and processes. We extend several typical SC handover algorithms for DC and compare them with the proposed algorithms in this study. Through numerical simulations, we evaluate the performance of these algorithms. The results demonstrate that the proposed algorithm minimizes unnecessary handovers, reduces service interruption time, and achieves the highest throughput.