Mobile edge computing (MEC) is considered as an effective solution to delay-sensitive services, and computing offloading, the central technology in MEC, can expand the capacity of resource-constrained mobile terminals (MTs). However, because of the interdependency among applications, and the dynamically changing and complex nature of the MEC environment, offloading decision making turns out to be an NP-hard problem. In the present work, a graph mapping offloading model (GMOM) based on deep reinforcement learning (DRL) is proposed to address the offloading problem of dependent tasks in MEC. Specifically, the MT application is first modeled into a directed acyclic graph (DAG), which is called a DAG task. Then, the DAG task is transformed into a subtask sequence vector according to the predefined order of priorities to facilitate processing. Finally, the sequence vector is input into an encoding-decoding framework based on the attention mechanism to obtain the offloading strategy vector. The GMOM is trained using the advanced proximal policy optimization (PPO) algorithm to minimize the comprehensive cost function including delay and energy consumption. Experiments show that the proposed model has good decision-making performance, with verified effectiveness in convergence, delay, and energy consumption.

Age of Information (AoI) is a metric to describe the timeliness of a system proposed in recent years. It measures the freshness of the latest received data from the perspective of the target node in the system. This work studies a kind of dynamic data acquisition system for urban security that can update and control the situation of urban environmental security by collecting environmental data. The collected data packets need to be uploaded to the cloud center in time for data update, which has high requirements on the timeliness of the system and freshness of data. However, due to the limited computing capacity of mobile terminals and the pressure of bandwidth for data transmission, problems such as high data execution delay and transmission interruption are caused. Emerging mobile edge computing (MEC), a new model of computing that extends cloud computing capabilities to the edge network, promises to solve these problems. This work focuses on the timeliness of the system, as measured by the average AoI across all mobile terminals. First, a timeliness optimization model is defined, and a multi-agent deep reinforcement learning (DRL) algorithm combined with an attention mechanism is proposed to carry out computing offloading and resource allocation through the continuous interaction between agent and environment; then, in order to improve algorithm performance and data security, the federated learning mode is proposed to train agents; finally, the proposed algorithm is compared with other main baseline algorithms based on deep reinforcement learning. The simulation results show that the proposed algorithm not only outperforms other algorithms in optimizing system timeliness, but also improves the stability of training.

No abstract