Distributed Congestion Marking in 5G UPF Nodes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication networks lack the ability to effectively indicate and mitigate congestion that occurs at nodes other than the Next Generation Radio Access Network gNodeB (NG-RAN gNB), as congestion marking is primarily based on conditions at the gNB, failing to account for congestion at other network nodes in the path.
Innovation Solution
Implementing congestion marking capabilities in devices along the communication path, including non-gNB devices and intermediate UPFs, which can detect and set congestion indicators in data packets, allowing for comprehensive congestion management across the network, even when congestion occurs at nodes other than the gNB.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If congestion marking is implemented only at NG-RAN gNB, then the system is simple to operate, but congestion information is incomplete and does not reflect congestion at other network nodes
Solution Approach 1:
The congestion marking function is segmented and distributed across multiple network nodes (gNB, UPFs, and other intermediate devices) rather than being centralized at a single location. Each node independently performs congestion detection and marking based on local conditions, enabling comprehensive congestion information collection while maintaining operational simplicity at each individual node.
Solution Approach 2:
Multiple types of network devices (gNB, UPFs, intermediate devices) are equipped with universal congestion marking capabilities. Each device type can perform the same core function of detecting local congestion and setting ECN bits, making the system adaptable to congestion at any location in the network path without requiring different mechanisms for different device types.
2Reliability
If congestion marking is performed at multiple network nodes, then complete congestion information is obtained, but the device complexity increases
Solution Approach 1:
Each network node performs self-service congestion detection and marking based on its own local buffer conditions and traffic state. Devices autonomously monitor their own congestion levels and set ECN bits in packets when experiencing congestion, without requiring complex coordination or control mechanisms between nodes. This self-service approach enhances reliability while keeping individual device complexity manageable.
Solution Approach 2:
Network nodes are pre-configured with congestion marking capabilities and ECN support. When a node experiences congestion, it can immediately mark packets without requiring real-time coordination or complex decision-making processes. The preliminary setup of marking rules and thresholds enables rapid response to congestion while maintaining operational simplicity.
3Productivity
If only gNB congestion is indicated, then the marking process is simple, but services requiring low latency cannot be properly optimized
Solution Approach 1:
End devices receive ECN-marked packets from any point in the network path and generate feedback to adjust their transmission rates accordingly. This feedback loop enables services requiring low latency to be optimized based on actual congestion conditions anywhere in the network, not just at the gNB. The complete congestion information from multiple nodes provides accurate feedback for rate adaptation and service optimization.
Data Source
AI summary
Data packet congestion marking is supported in devices which may form part of a communications path. The devices which can perform marking can, and in some embodiments do, include non-gNB devices and non-anchor point UPFs, e.g., devices other than gNBs and PSA UPFs such as intermediate UPFs, N3IWFs, TNGFs, and W-AGFs, in addition to gNBs and anchor UPFs. Intermediate UPFs corresponding to a communications session will set the congestion indicator value in a data packet's IP header, of a data packet passing through the UPF, to indicate congestion, if it is not already set to indicate congestion when congestion is detected at the intermediate UPF. A session management function in some embodiments controls which devices perform congestion marking and enables/disables congestion marking, e.g., on a per communications session basis while the session is ongoing and/or at various times during a session establishment or device registration process.