Delay-Sensitive Aggregation Coded Repair: Towards Low-Energy LEO Storage Constellation

Zexi Li,Shushi Gu,Zhikai Zhang,Qinyu Zhang,Wei Xiang
DOI: https://doi.org/10.1109/wcnc57260.2024.10570915
2024-01-01
Abstract:The LEO storage constellation (LSC) has gained significant popularity thanks to its inter-satellite massive con-nectivity and space-terrestrial integrated data storage capability. However, the satellite disks will fail occasionally due to exhausting energy or space debris, which leads to data loss and network unavailable. In terrestrial data centers, aggregation coded repair (ACR) is an emerging data repair technique, which can reduce repair traffic by aggregating source data flows on intermediate nodes. However, existing ACR research does not focus on the problems of large propagation distance and energy consumption constraints in LSC. In this paper, we propose the coding path tree (CPT) model, which is a staged tree graph containing aggregation vectors and distance-related edge weights. For reducing repair delay and energy cost, we further propose the Delay-Sensitive Energy-Efficient ACR (DE-ACR) scheme, which is based on a CPT construction algorithm that combines the ideas of the shortest path and minimum Steiner tree. System-level experiment demonstrates that DE-ACR obtains a 7% reduction in repair delay compared to the single repair pattern tree (SRPT), and achieves a 55 % decrease in energy consumption compared to the shortest path ACR (SP-ACR) in LSC.
What problem does this paper attempt to address?