GNAT Traffic Control Using DRL for Multi-SATCOM QoS
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current SATCOM systems face challenges in managing network traffic across multiple satellite networks with varying connection properties, necessitating sophisticated traffic management to meet Quality of Service (QoS) requirements for high throughput and low latency.
Innovation Solution
A deep reinforcement learning (DRL)-based dynamic network traffic management (DNTM) system utilizing a GNAT controller, LAN router, network switch, and WAN routers to measure traffic states, adjust bandwidth assignments, and optimize bandwidth allocation using a DNN to predict long-term rewards and satisfy QoS requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple SATCOM systems are utilized simultaneously through different satellite networks, then network throughput and reliability are improved, but traffic management complexity increases due to large differences in connection properties
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting bandwidth allocation parameters based on real-time traffic states and QoS requirements. The DRL model continuously learns and adapts bandwidth assignment parameters to optimize performance across heterogeneous SATCOM networks, resolving the contradiction by making the system adaptable rather than static.
Solution Approach 2:
The system implements self-service through autonomous DRL-based decision-making that automatically manages traffic across multiple SATCOM networks without manual intervention. The controller independently monitors traffic states, evaluates QoS requirements, and adjusts bandwidth allocation, eliminating the need for complex manual traffic management while improving reliability.
2Productivity
If bandwidth is dynamically adjusted to meet QoS requirements, then communication throughput and latency are improved, but computational complexity for traffic management increases
Solution Approach 1:
The patent replaces traditional mechanical/manual traffic management mechanisms with an intelligent DRL-based system. Instead of rule-based or manual bandwidth allocation, the system uses machine learning models that automatically learn optimal bandwidth assignment strategies, reducing computational overhead while improving throughput and latency performance.
Solution Approach 2:
The system implements continuous feedback loops where the DRL controller monitors traffic states, evaluates QoS metrics, and adjusts bandwidth allocation in real-time. This feedback mechanism enables the system to adapt to changing network conditions dynamically, improving communication throughput while maintaining manageable computational complexity through iterative learning.
3Reliability
If deep reinforcement learning model is used for bandwidth assignment, then long-term reward and QoS satisfaction are improved, but training and computation time increase
Solution Approach 1:
The patent applies preliminary action by pre-training the DRL model offline to learn optimal bandwidth assignment strategies before deployment. The model is trained on historical traffic data and QoS requirements in advance, so that during runtime, it can quickly make decisions without extensive real-time computation, reducing training and computation time while maintaining high QoS satisfaction.
Data Source
AI summary
The present disclosure provides a deep reinforcement learning (DRL) based dynamic network traffic management system including a LAN router, a plurality of WAN routers, a network switch, and a GNAT controller configured to measure one or more traffic states of a plurality of data flows, obtain an expected reward at the current time point, obtain the one or more traffic states to input to a DRL model to provide an expected reward of each data flow estimated for a next time point, obtain a target reward at the current time point, adjust parameters of the DRL model, predict a plurality of long-term rewards using the trained DRL model, select one of the plurality of long-term rewards, and adjust the bandwidth assigned to each data flow based on the selected long-term reward.