Deep Learning for Hybrid 5G Services in Mobile Edge Computing Systems: Learn From a Digital Twin

In this paper, we consider a mobile edge computing system with both ultra-reliable and low-latency communications services and delay tolerant services. We aim to minimize the normalized energy consumption, defined as the energy consumption per bit, by optimizing user association, resource allocation, and offloading probabilities subject to the quality-of-service requirements. The user association is managed by the mobility management entity (MME), while resource allocation and offloading probabilities are determined by each access point (AP). We propose a deep learning (DL) architecture, where a digital twin of the real network environment is used to train the DL algorithm off-line at a central server. From the pre-trained deep neural network (DNN), the MME can obtain user association scheme in a real-time manner. Considering that the real networks are not static, the digital twin monitors the variation of real networks and updates the DNN accordingly. For a given user association scheme, we propose an optimization algorithm to find the optimal resource allocation and offloading probabilities at each AP. The simulation results show that our method can achieve lower normalized energy consumption with less computation complexity compared with an existing method and approach to the performance of the global optimal solution.

I. Introduction

In the 5th generation (5G) communication systems, there are diverse applications ranging from high data rate delay tolerant services to ultra-reliable and low-latency communications (URLLC) [1]. By achieving ultra-low end-to-end (E2E) delay and ultra-high reliability, URLLC lies the foundation for emerging latency-critical applications, such as factory automation, autonomous vehicles, and virtual/augmented reality [2]. Devices in these applications will generate some tasks that require to be processed within a short time. To reduce processing time at the local server of each device and to avoid delays in backhauls and core networks, mobile edge computing (MEC) is one promising solution [3]. However, when a task is packetized in a short packet and offloaded to a MEC server via a wireless link, the packet may be lost when the channel is in deep fading [4]. Besides, short blocklength channel codes will cause the none-zero decoding error rate, even for an arbitrarily high signal-to-noise ratio (SNR) [5], [6]. Thus, achieving ultra-high reliability and ultra-low latency is very challenging in MEC systems.