Using Virtual Network Architecture as the foundation, Space-Air-Ground Integrated Multi-domain Network Resource Orchestration is a DRL Method.

PROJECT TITLE :

Space-Air-Ground Integrated Multi-domain Network Resource Orchestration based on Virtual Network Architecture a DRL Method

ABSTRACT:

Due to deployment, coverage, and capacity issues, traditional ground wireless Communication networks are unable to provide high-quality services for artificial intelligence (AI) applications such as intelligent transportation systems (ITS). The space-air-ground integrated network, also known as SAGIN, has emerged as a central topic of investigation in the sector. SAGIN is more flexible and reliable than traditional wireless Communication networks. It also has wider coverage and a higher quality of seamless connection than other networks. However, the deployment and use of SAGIN still faces enormous challenges as a result of its inherent heterogeneity, time-varying, and self-organizing characteristics. One of these challenges is the orchestration of heterogeneous resources, which is an important issue. We model SAGIN's heterogeneous resource orchestration as a multi-domain virtual network embedding (VNE) problem using virtual network architecture and deep reinforcement learning (DRL), and we propose a SAGIN cross-domain virtual network embedding (VNE) algorithm. Both of these models are based on virtual network architecture. We model the various network segments that make up SAGIN, and we configure the network attributes in accordance with the current state of SAGIN and the requirements of the users. A five-layer policy network is responsible for acting as the agent in DRL. We construct a feature matrix by making use of network attributes that have been extracted from SAGIN, and then we use this matrix as the environment for training agents. It is possible, through training, to derive the probability of each underlying node being embedded in the tree. During the test phase, we will finish the embedding process of virtual nodes and links one at a time, based on the probability that each will be used. In the end, we evaluate the performance of the algorithm based on both the training data and the test data.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here