EFS: Efficient Storage Optimization for Multistage Flow-Table in Software-Defined Satellite Network

Software-Defined Satellite Network (SDSN) plays an essential role in future networks. With the rapid development of networks and the constant enrichment of services, the number of filter rules in flow tables becomes enormous, which challenges the limited resources of on-board switches. For the memory shortage of flow tables in SDSN, we propose an expanded-field search (EFS) algorithm, which supports storage compression during both the initialization and update of multistage flow-table. EFS considers the cost distinction between the static random-access memory (SRAM) and the ternary content addressable memory (TCAM). It applies a novel search strategy to expand the search field from $O(N)$ to $O(N^{2})$ for better convergence, and employs a statistical inference method to simplify the computation. Simulation results show that EFS takes a short run time and has a storage compression close to the global optimum, which outperforms existing algorithms significantly. In addition, the EFS also shows better performance in terms of search cost.