Significance of Task Information Optimization for UAV-based Real-Time Target Tracking

With the advent of 6G technology and artificial intelligence (AI), the space-air-ground-sea integrated network (SAGSIN) aims to deliver ubiquitous connectivity, overcoming communication coverage challenges in remote areas. However, the high dynamics and heterogeneity of such networks pose significant challenges for the timely and efficient transmission of information. Traditional communication, focusing on bit-level reconstruction, often overlooks the semantic richness and effectiveness of information, which are essential for SAGSIN. To address these issues, this paper introduces a closed-loop Unmanned Aerial Vehicle (UAV) target tracking system. A High-Altitude Platform (HAP) observes time-sensitive targets and provides status updates to the UAV, enhancing its tracking capabilities through combined observations and assistance. We propose the Significance of Task Information (SoTI) metric to quantify the effectiveness of tracking tasks at a task-oriented level. Considering the HAP’s transmission energy consumption, we formulate a task-oriented transmission problem, which minimizes the total task cost, weighted by SoTI and energy consumption. This problem is modeled as a Markov Decision Process (MDP) and efficiently solved using the Deep Q-Network (DQN) algorithm. Numerical results show that our SoTI-based transmission policy reduces task costs while balancing transmission energy consumption and task completion time across varying target status change rates.