A Novel Space-Based Hosting Approach for Ultra Low Latency Web Services

Abstract

The lack of terrestrial 4G and 5G communication infrastructure in rural and remote areas aggravates the digital divide. Satellite communication is important to achieve the universal goal of connecting the digitally unconnected population through global coverage. In this paper, first, we discuss the feasibility of hosting web services in satellites through several examples of radiation-hardened processors and computational boards. Next, we propose a Space-Based Hosting Service (SBHS) approach to deploy the content-servers or copies of terrestrial content-servers in satellites to achieve the low-latency service requirements. For SBHS, we develop mathematical models for communication, queuing, as well as computation, and use B+ tree to estimate computational delay and computational energy consumption of the content-server. Further, an optimization algorithm using Markov Decision Process (MDP) is designed to schedule computational cores in Low Earth Orbit (LEO) satellites to minimize queuing delay and computational energy consumption. The efficacy of our SBHS is evaluated using extensive simulations with the Wikipedia use-case in the Iridium-NEXT satellite constellation model under varying traffic conditions from different countries. The results show that SHBS is able to achieve less than 10 ms round trip latency worldwide and minimum end-to-end delays of 9.19 ms and 24.29 ms, respectively, for text-based and multimedia-based traffic. The 92.51% and 88.66% reduction in delay for text and multimedia traffic, respectively, compared to the benchmark schemes underlines the benefits of hosting content-servers in space for achieving ultra-low latency.

Keywords

• LEO satellite
• B+ tree
• content server
• Markov decision process (MDP)
• satellite networks
• Wikipedia