SDN Controller Load Balancing via Non-Zero Sum Game
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for optimizing controller placement in Software Defined Networks (SDNs) fail to dynamically adjust to varying network loads, leading to suboptimal performance, underutilization, and increased operational costs due to static mapping and reliance on centralized control entities, which can result in poor Quality-of-Service and excessive resource consumption.
Innovation Solution
Implementing a non-zero sum game based network control plane optimization operation that allows each controller to dynamically compute its payoff value and adjust its load distribution, enabling dynamic addition or deletion of controllers and optimal placement to balance utilization and minimize delays, independent of centralized control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a centralized control entity is used to manage controller placement in SDN, then control decisions can be made centrally, but the system becomes less adaptable to varying network loads and incurs higher operational costs
Solution Approach 1:
The patent segments the centralized control function into distributed autonomous controllers, each capable of independent decision-making. Controllers are divided into different types (ingress, egress, transit) that can autonomously manage their respective segments of the network, eliminating the single-point bottleneck of centralized control while maintaining coordinated operation through peer-to-peer communication.
Solution Approach 2:
The patent implements dynamic controller placement and load balancing mechanisms where controllers can be added, removed, or migrated based on real-time network conditions. The system dynamically adjusts controller utilization levels and redistributes network device assignments when load thresholds are exceeded, enabling continuous adaptation to varying network demands without centralized intervention.
2Productivity
If static controller placement is used in SDN, then implementation is simpler, but performance becomes suboptimal under varying network loads
Solution Approach 1:
The patent enables controllers to autonomously monitor their own load conditions and trigger load balancing operations when thresholds are exceeded. Controllers self-evaluate their utilization metrics, automatically initiate redistribution of network devices to underutilized controllers, and adjust their operational state without external control, thereby optimizing performance through self-service mechanisms.
Solution Approach 2:
The patent implements continuous feedback loops where controllers monitor their own performance metrics and network conditions, then adjust their behavior accordingly. Load balancing decisions are driven by real-time feedback on controller utilization, network device distribution, and traffic patterns, enabling automated optimization that adapts to changing conditions while maintaining implementation simplicity through rule-based decision logic.
3Reliability
If controllers continuously monitor and adjust load distribution, then optimal performance is achieved, but computational overhead and resource consumption increase
Solution Approach 1:
The patent implements periodic load monitoring and threshold-based triggering mechanisms rather than continuous real-time analysis. Controllers evaluate their load conditions at defined intervals and initiate load balancing operations only when predetermined thresholds are exceeded, reducing computational overhead while maintaining reliable QoS through periodic assessments of network health and performance metrics.
Solution Approach 2:
The patent dynamically adjusts monitoring and evaluation parameters based on network conditions, transitioning between different levels of scrutiny and computational intensity. During normal operating conditions, controllers use simplified threshold-based decisions with lower computational requirements, while automatically increasing monitoring frequency and analysis depth only when performance degradation is detected, thereby optimizing resource consumption relative to actual network needs.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Optimization of control plane in a software defined network includes obtaining peer information of at least one neighbouring network controller (110) by a network controller (110) and determining a traffic profile variation. The method further includes computing of a self payoff value indicative of one of optimum utilization, underutilization and overutilization of the network controller (110). The method further includes initiating a non-zero sum game based network control plane optimization operation based on the self payoff value and the traffic profile of the neighbouring network controllers, and may include one of activating additional network controller(s) (110), transferring control of one or more network devices (106) managed by the network controller(s) (110) to a neighbouring greedy network controller (110), deactivating the network controller (110), and transferring control of one or more additional network devices (106) managed by the neighbouring network controller(s) to the greedy network controller (110).