Lo, Chi Jen RogerChi Jen RogerLoHUNG-YUN HSIEH2023-07-172023-07-172022-01-019781665486729https://scholars.lib.ntu.edu.tw/handle/123456789/633738In this paper, we address the problem of minimizing the queue length for data gathering in wireless sensor networks (WSNs) while considering the quality of the information received at the sink. We extend the nominated Slepian-Wolf-Cover bound for distributed source coding (DSC) into a blockwise streaming scheme, where distributed sources could be compressed without loss exploiting the spatio-temporal correlations between samples from individual sensor nodes. Moreover, unlike the classic bound which requires statistical information of the correlated sources a priori, our novel time-average inequality can dynamically adapt to arbitrary ergodic statistics for every scheduling period. On the other hand, since radio resource is often finite, not all information stored in the queues can be delivered to the base station. As long as the desired fidelity requirement is satisfied, we find the queues can be further shortened by proactively discarding some data within. We contrive a stochastic online optimization problem and resolve it with the proposed information-centric scheduling algorithm. Evaluation results show such an information-centric design can significantly decrease the time-averaged queue length.conference paper10.1109/SDF55338.2022.99319512-s2.0-85142441680https://api.elsevier.com/content/abstract/scopus_id/85142441680