Timely Gossiping for Semantic-Based Status Tracking in Robotic Communications

Moving beyond interconnected devices, recent advancements in artificial intelligence (AI) technologies have fed the vision of networked robotic systems. The collaboration among robots relies on timely exchange of status updates among themselves through decentralized communication mechanisms, such as the gossip protocol. Considering the limited transmission link capacity between two robots, each robot could leverage its local computing capability to extract semantic features from high-dimensional sensing data to reduce the communication cost. In this sense, in order to improve the timeliness of status update exchange, both computation and communication stages should be taken into consideration for gossip protocol design. In this paper, we derive the closed-form expression of timeliness of status updates (ToS) in hierarchical gossip networks, which explicitly characterizes the coupling among semantic compression ratio, computing rate at the source robot, and link capacity between robots. Moreover, we exploit the inherent temporal dynamics of sensing data to evaluate the process-related timeliness of status updates, which further characterizes the impact of the underlying structure of sensing data on the timeliness of status updates. The obtained analytical formulas can serve as novel performance metrics to design the gossip protocol for timely robotic communications. The feasibility and accuracy of the theoretical derivation are validated. Numerical results demonstrate the non-monotonic relationship between ToS and semantic compression ratio, which reveals the necessity of finding an optimal combination of computation and transmission parameters. An illustrative case study of vehicle speed tracking shows a tight correlation between the average tracking error and ToS, validating the practical meanings of the proposed metrics.