A Task Offloading Algorithm based on Deep Reinforcement Learning with BiLSTM for Mobile Edge Computing

Task offloading in mobile edge computing(MEC) environments is an important research topic for enhancing user experience. This study proposes a novel approach to optimize task offloading in MEC based on deep reinforcement learning(DRL). This study replaces the LSTM(Long Short-Term Memory) in the existing QOCO(QoE-Oriented Computation Offloading) algorithm with BiLSTM(Bidirectional LSTM). The proposed method uses BiLSTM to improve the capture of dynamic workloads in edge nodes. This results in a performance enhancement of approximately 8% reduction in drop rate and a 6% reduction in the number of dropped tasks, as confirmed through experiments. These results show the advantages of combining DRL and BiLSTM in optimizing task offloading in MEC environments.