Entry Aggregation and Early Match Using Hidden Markov Model of Flow Table in SDN

Area of Science:

• Computer Science
• Networking

Background:

• Software-Defined Networking (SDN) utilizes multiple flow tables (MFT) for enhanced flexibility.
• MFT size is constrained by expensive Ternary Content Addressable Memory (TCAM), and pipeline mechanisms lead to long flow processing times.

Purpose of the Study:

• To propose a novel approach, Agg-ExTable, for efficient management of MFT in SDN.
• To address the limitations of MFT size and flow processing time.

Main Methods:

• Flow entries in MFT are periodically aggregated using pruning and the Quine-McCluskey algorithm.
• A front-end ExTable is created using the saved memory to store popular flow entries for early matching.
• Popular entries are identified using a Hidden Markov model based on match frequency and probability.

Main Results:

• The Agg-ExTable scheme achieves approximately 45% space saving for MFT.
• The proposed scheme significantly decreases flow processing time compared to existing methods.

Conclusions:

• Agg-ExTable offers an efficient solution for managing MFT in SDN environments.
• The approach effectively overcomes TCAM limitations and reduces processing latency.