Data Freshness Oriented Sampling Under Unknown Delay Statistics
Haoyue Tang
Abstract:
We will revisit data-freshness oriented sampling problem for a transmitter-receiver pair with random communication delay. The goal is to design a sampling strategy in the absence of delay distribution statistics. When the evoluation of the time-varying source is unavailable to the sampler, we use the Age of Information, namely the time elapsed since the latest information has been generated, to capture information freshness at the receiver. We propose an online algorithm that minimizes the AoI performance. We prove that the algorithm learns the optimum strategy almost surely, and the learning rate is minimax order optimal. Next, we turn to sampling problem for the Wiener process. We propose online algorithm that minimizes the MSE and show its optimality.