Proactive eviction of flow entry for SDN based on hidden Markov model

Ullah

et al. 2023

Preprint

Self Cite

Software-Defined Networking (SDN) has emerged as an innovative method of networking that offers effective Nanyang Institute of Technology management and remarkable flexibility. However, current SDN-based solutions address networks' static nature or concentrate primarily on backbone networks, where network dynamics have minimal impact. This research presents a new approach, specifically the mobility-aware adaptive flow entry placement scheme for SDN-based Internet of Things (IoT) environments, to address the mobility aspect of networks. The proposed scheme employs the Q-learning algorithm to predict the next possible location of end devices, while the cost-sensitive AdaBoost algorithm is utilized to select heavy and active flows. Consequently, efficient flow rules for incoming flows can be dynamically implemented without the need for intervention from the controllers. Extensive computer simulations demonstrate that this approach significantly enhances match probability and prediction accuracy, while concurrently reducing the number of table misses and resource expenditure compared to existing schemes.

Section: A Flow Evictionmentioning

confidence: 99%

Predictive Mobility and Cost-Aware Flow Placement in SDN-Based IoT Networks: A Q-Learning Approach

Ullah

et al. 2023

Preprint

Self Cite

“…Some of them use machine learning models as well to predict the rules for eviction. One of the first ever works that use historical data for eviction in SDN flow tables is discussed in Huang et al 19 Another earliest technique named Flow Master was proposed by Kannan and Banerjee 20 that uses Markov model‐based prediction to pro‐actively evict dead flows. A hardware prototype was implemented in the NetFPGA platform.…”

Section: Background and Motivationmentioning

confidence: 99%

A novel eviction policy based on shortest remaining time for software defined networking flow tables

Dhandapani,

Thanganadar Thangathai,

Hamead Haja Moinudeen

2023

Int J Network Mgmt

SummarySoftware defined networking is a modern paradigm that divides the control plane from the data plane for improved network manageability. A flow table in the data plane has limited and expensive memory called TCAM. The presence of unwanted flow rules would lead to flow bloat conditions and make the lookup operation inefficient. Eviction schemes based on LRU policy have been widely studied in the literature which preempts the life of the least recently used flow rule and reduces the occupancy of the flow table. LRU considers the past behavior for the eviction of a flow rule. This paper proposes a novel policy that preempts a flow rule by considering its future characteristic, the shortest remaining time (SRT). A rule with a higher probability of being used is avoided from eviction to mitigate the degradation of performance. The modeling of the SRT technique exhibits better utilization of a flow rule that has higher probability of being used. On the other hand, LRU does not guarantee that the evicted flow rule will not be used frequently in the future and it has been shown that an incorrectly evicted flow rule incurs controller delay. The experimental results show that for different traffic rates, SRT has reduced the delay by 15%, reinstallation count by 25%, and jitter by 40%. SRT has increased utilization by 22% compared to LRU.

“…Because our network is softwarebased, switches cannot manipulate it in case of high overload which is the same as the aggregation, and they usually operate on broadcasting principles. The base paper used a tree pattern and on the switches in the ow tables, receiving, storing and sending messages used a hierarchical pattern [10], [11].…”

Section: Introductionmentioning

confidence: 99%

Improving Hierarchical Traffic Engineering and Reducing the Congestion with PSO in SDN and DataCenter Networks

Azizi

Aghaei

Ghaffarian

et al. 2023

Preprint

Software-defined networks separate the controlling and data layers in computer networks. It will enable central management and the ability to define policies and apply them to network hardware. Traffic engineering, which measures and manages network traffic, is a major activity in these networks. Data should be monitored when traffic is excessive and traffic should be manipulated. Among the most important issues to consider in a network is monitoring. In this case, the network is based on SDN. Overflows on a network do not allow switches to manipulate the network. As mentioned above, this overflow is aggregation and usually based on broadcasting principles. This article describes how links work at the network level using parameters of flow, link capacity, and the optimal allocation of profits. This paper attempts to reduce congestion, reduce packet losses, and optimize delay through a number of methods. This paper proposes a method (PSOCB) based on SDN. Using the meta-heuristic methods to allocate the cluster switches existing in the network between the source and destination and the method of optimizing, the throughput of particles is increased by 13% in switches, and 8% in links. There is a reduction in delay when compared to the base paper. The proposed method experienced congestion at one time higher than the base paper, and at other times lower. In case of congestion removal, the network returns to its normal state. Overall, this method improves congestion as well.