Freshness-Aware Energy Efficiency Optimization for Integrated Access and Backhaul Networks

Age of information (AoI) has introduced a new dimension into the design of real-time monitoring networks. However, how to achieve energy-efficient freshness-aware information transmission with energy-constrained Internet of Things (IoT) devices remains a challenge. To solve this challenge, in this paper, we consider a new metric called freshness-aware energy efficiency (FAEE), which is defined as the ratio of AoI improvement to transmission energy consumption. We study a freshness-aware integrated access and backhaul (IAB) network in which multiple IoT devices (IDs) that are accountable for generating and transmitting status updates frequently over time so as to maintain the freshness of information observed at the IAB donor. For this system setup, we first formulate three mixed discrete-continuous optimization problems in order to maximize the long-term average FAEE under different transmission scheduling schemes, including round-robin (RR), sub-channel allocation (SA), and time-resource allocation (TA). By constructing average-reward Markov decision processes (MDPs) with mixed discrete-continuous action space to model these optimization problems, we propose a novel proximal policy optimization (PPO) based deep reinforcement learning (DRL) framework, which is referred to as APO-CD, to learn sub-optimal policies for status update strategy and resource allocation. Extensive simulation results are provided to show the effectiveness of our proposed algorithm and to reveal several design insights of freshness-aware IAB networks.