Dynamically optimizing routing within a decentralized network

A system comprising:

• a node of a decentralized network that is layered on top of a blockchain network, the node configured with settings that control routing of network traffic through the node of the decentralized network;
• and
• a node operator associated with the node and configured to:
• obtain state data describing a state of the decentralized network;
• process the state data using one or more machine learning models to generate a prediction of optimized settings for the node, wherein:
• the one or more machine learning models are used to perform multiple simulations of traffic flows for a period of time over the decentralized network with the state data;
• a state associated with the node is adjusted for each of the multiple simulations, and
• the prediction of the optimized settings for the node is based on optimized routing within the decentralized network in the multiple simulations;
• and
• dynamically adjust one or more of the settings of the node to conform to the optimized settings.

Computer-readable storage media storing instructions that are executable by one or more processors to perform operations comprising:

• accessing historical state data to train one or more models using machine learning, wherein the historical state data is associated with a decentralized network layered on top of a blockchain network, and wherein the one or more models are trained to optimize routing within the decentralized network;
• monitoring, in near-real-time and using the one or more models, state data associated with the decentralized network;
• performing multiple simulations of traffic flows for a period of time over the decentralized network using the state data and the one or more models, wherein a state associated with a node of the decentralized network is adjusted for each of the multiple simulations;
• determining an optimized state of the node to optimize routing within the decentralized network based on the multiple simulations;
• and
• dynamically causing one or more actions to be performed by a node operator that configure the node to operate in the optimized state.