Risk-Aware Task Offloading and Resource Allocation in FiWi HetNets

Abstract

Multimedia service, as one of the most important business in Internet of Things (IoT), poses a great challenge to network throughput and reliable service. MEC enhanced fiber-wireless heterogeneous network (FiWi HetNets) is considered to be a cost-effective scheme to improve the network coverage of IoT. However, due to the communication uncertainty caused by the change of channel conditions and the limited resources of the edge server, the offloaded task is at risk of failure caused by packet loss or overload of the edge server. Therefore, this paper proposes a risk-aware computing offload and resource allocation method to ensure the QoS of multimedia services. Firstly, we design a risk perception model based on conditional value at risk (CVaR) to measure the risk of 5G and WiFi wireless transmission modes, and formulate the optimization problem. Then, an A3C based optimization method for wireless access, computation offloading and resource allocation is proposed, to balance the end-to-end service delay. Simulation results verify the reliability and superiority of this method.