I. Introduction
Compared with traditional terrestrial networks, space information networks (SINs) have many advantages, such as wide coverage and flexible networking [1]–[4]. Therefore, SINs have attracted intensive research interests [1]–[8]. The SINs usually consist of different types of space platforms, such as geostationary Earth orbit (GEO) satellites, middle Earth orbit (MEO) satellites, low Earth orbit (LEO) satellites, stratospheric balloons, manned and unmanned aircraft, and so on [1]–[3]. Furthermore, the topology of the SINs is dynamic but predictable [8]–[11]. Therefore, the time-varying graphs, such as snapshot graph [9], time-aggregated graph (TAG) [9], [10], [12], and time-expanded graph (TEG) [11], [13], [14], have been widely used to model SINs in order to describe their time-varying characteristics. The snapshot graph uses a series of snapshots to describe the dynamic evolution of the SINs [9]. However, the snapshot graph cannot exploit the storage resource of the node in SINs because there are no connections between snapshots [9], [12]. In contrast, the TEG was proposed to connect the two consecutive snapshots of the same node by caching links, which facilitates the exploitation of the storage of the nodes in the SINs [9], [12].