Abstract:

Due to the constantly changing positions of satellites， UAVs and ground stations， the space space edge computing network link is not fixed， and the network needs to respond quickly to user requests， which requires high throughput and real-time， increasing the difficulty of network resource allocation. In this regard， this research proposes a network resource allocation method for space edge computing based on deep Q network algorithm. Firstly， considering the dynamic nature of network topology and resource heterogeneity， establish a communication model between resources to provide a basic framework for resource allocation； Then， based on the maximum throughput， a resource allocation objective function is designed， and a Markov decision model is used to express the objective function. The resource allocation problem is transformed into a sequential decision problem， which facilitates decision-making in a dynamically changing network environment； Finally， based on the deep Q-network algorithm， the objective function is solved， and through reinforcement learning， the algorithm can learn the optimal resource allocation strategy through interaction with the environment， adapting to the real-time and dynamic nature of the network. The experimental results show that after applying this method， the cumulative return of the network is higher and the average energy consumption of resource tasks is reduced， indicating that this method is more suitable for practical applications.

Key words: integrated network of air， space， and space, deep Q-network algorithm, edge computing, resource allocation, markov decision model