Luo Hongqiu, Yan Tingting, Hu Shengbo
Institute of Intelligent Information Processing, Guizhou Normal University, Guiyang 550001, China.
School of Computer and Electronic Information, Guizhou Qiannan College of Science and Technology, Guiyang 550600, China.
Sensors (Basel). 2024 May 25;24(11):3412. doi: 10.3390/s24113412.
Information-Centric Networking (ICN) is the emerging next-generation internet paradigm. The Low Earth Orbit (LEO) satellite mega-constellation based on ICN can achieve seamless global coverage and provide excellent support for Internet of Things (IoT) services. Additionally, in-network caching, typically characteristic of ICN, plays a paramount role in network performance. Therefore, the in-network caching policy is one of the hotspot problems. Especially, compared to caching traditional internet content, in-networking caching IoT content is more challenging, since the IoT content lifetime is small and transient. In this paper, firstly, the framework of the LEO satellite mega-constellation Information-Centric Networking for IoT (LEO-SMC-ICN-IoT) is proposed. Then, introducing the concept of "viscosity", the proposed Caching Algorithm based on the Random Forest (CARF) policy of satellite nodes combines both content popularity prediction and satellite nodes location prediction, for achieving good cache matching between the satellite nodes and content. And using the matching rule, the Random Forest (RF) algorithm is adopted to predict the matching relationship among satellite nodes and content for guiding the deployment of caches. Especially, the content is cached in advance at the future satellite to maintain communication with the current ground segment at the time of satellite switchover. Additionally, the policy considers both the IoT content lifetime and the freshness. Finally, a simulation platform with LEO satellite mega-constellation based on ICN is developed in Network Simulator 3 (NS-3). The simulation results show that the proposed caching policy compared with the Leave Copy Everywhere (LCE), the opportunistic (OPP), the Leave Copy down (LCD), and the probabilistic algorithm which caches each content with probability 0.5 (prob 0.5) yield a significant performance improvement, such as the average number of hops, i.e., delay, cache hit rate, and throughput.
以信息为中心的网络（ICN）是新兴的下一代互联网范式。基于ICN的低地球轨道（LEO）卫星巨型星座能够实现全球无缝覆盖，并为物联网（IoT）服务提供出色支持。此外，作为ICN典型特征的网络内缓存对网络性能起着至关重要的作用。因此，网络内缓存策略是热点问题之一。特别是，与缓存传统互联网内容相比，网络内缓存物联网内容更具挑战性，因为物联网内容的生命周期短且具有瞬态性。本文首先提出了用于物联网的LEO卫星巨型星座以信息为中心的网络（LEO-SMC-ICN-IoT）框架。然后，引入“粘性”概念，所提出的基于卫星节点随机森林（CARF）策略的缓存算法结合了内容流行度预测和卫星节点位置预测，以实现卫星节点与内容之间良好的缓存匹配。并使用匹配规则，采用随机森林（RF）算法预测卫星节点与内容之间的匹配关系，以指导缓存的部署。特别是，内容会提前缓存在未来的卫星上，以便在卫星切换时与当前地面段保持通信。此外，该策略同时考虑了物联网内容的生命周期和新鲜度。最后，在网络模拟器3（NS-3）中开发了基于ICN的LEO卫星巨型星座仿真平台。仿真结果表明，与随处留存副本（LCE）、机会主义（OPP）、向下留存副本（LCD）以及以概率0.5缓存每个内容的概率算法（prob 0.5）相比，所提出的缓存策略在平均跳数（即延迟）、缓存命中率和吞吐量等方面有显著的性能提升。
