A Service Function Chain Deployment Scheme of the Software Defined Satellite Network

Abstract

The aim of this study was to reduce the influence of highly dynamic substrate network topology and large transmission delay caused by long communication distance between satellites and ground in Software Defined Satellite Networks (SDSNs), and a Partial Observation Markov Decision Process (POMDP)-based service function chain (SFC) deployment scheme of SDSN is proposed. Under this SDSN architecture, the network topology changes could be obtained through the SDN centralized control ability. Due to the topology changes, which may cause inevitable observation errors and transmission delays, the complete actual topology and network states cannot be obtained in real time. Thus, we put forward a POMDP model–based SFC deployment scheme, and an approximate iterative algorithm to solve the problem, aiming at optimizing the end-to-end network delay in the SDSN. The simulation results show that our model can optimize the delay of SFC deployment process, and improve the resource utilization and network throughput of the SDSN.

A Satellite Network Service Function Chain Orchestration Method Based on Node Stability

“…In reference [19], the authors presented a goal-weighted path selection algorithm for optimizing SFC traffic routing in multi-layer satellite networks. Reference [20] employed a Hidden Markov Decision Model to address challenges related to satellite network dynamics and significant inter-satellite link delays. In terms of SFC reconfiguration in satellite networks, reference [21] utilized a tabu search-based algorithm to maximize the cumulative profit of service providers by readjusting existing SFCs and deploying new SFC requests.…”

“…There are two available specifications for both CPU and memory, namely, {48core, 64core} and {256M b, 512M b}. Storage follows a discrete uniform distributionU (160, 320)M b, link bandwidth follows a discrete uniform distributionB ∼ U (180, 240), the source and destination nodes of the SFC are randomly distributed in 14 different geographical locations within China, the VNF number of each SFC are following a discrete uniform distributionU (3,5), the traffic demands follow a discrete uniform distributionU (5,20), the resource consumption coefficients also follow discrete uniform distribution can be expressed as coef f cpu ∼ U (1, 3)/5M bps, coef f memoey ∼ U (0, 4)/5M bps, and coef f storage ∼ U (0, 5)/5M bps, the duration and maximum allowable latency can be expressed as Duration ∼ U (5, 20), Delay ∼ U (25, 35), The above parameters can be summarized in the following table.…”