Performance Analysis And Optimization Of Non-orthogonal Multiple Access Technology For Fog Radio Access Networks

The diversified user services have brought new challenges to the current wireless network architecture.In traditional centralized wireless access networks,the transmission latency is relatively high and the fronthaul and backhaul links are capacity-limited,leading to the constrained overall network performance.Fog radio access networks(F-RANs),which deploy part of the signal processing functionalities at the cloud processors,can improve the performance of interference suppression.In addition,F-RANs caches some contents to the network edge in advance,thus can alleviate the load of fronthaul links and the cloud processors,and reduce the transmission latency.Non-orthogonal multiple access(NOMA)technology,which uses non-orthogonal resources to distinguish different users,increases the number of users on the premise of guaranteeing the quality of service(QoS).As such,integrating NOMA technology into the physical layer of F-RANs can further increase the network capacity,reduce the transmission latency,and increase the number of supported users.In this thesis,the theoretical performance of NOMA based F-RANs is modeled.Based on which,the impacts of key parameters on the network performance is analyzed.Moreover,the access modes of users in the network are dynamically optimized.The main contents and contributions of this thesis are summarized as follows:1.To investigate the performance of NOMA-based F-RANs,the performance analysis model of NOMA-based F-RANs is proposed to research the theoretical performance of downlink transmission scenario.In this scenario,a typical user suffers from the interference from multiple transmission nodes as well as the inter-user interference,making it complicated to theoretically analyze the network performance.Using tools from stochastic geometry,the closed-form expressions of coverage probability and area spectrum efficiency are derived,and the key parameters that can impact the network performance are identified.The Monte Carlo simulations verify the correctness of the theoretical analysis.In addition,the specific impacts of key parameters on the network performance and the applicability of NOMA technology are examined.Specifically,the simulation results show that NOMA transmission technology can improve the spectrum efficiency by up to 76%compared with traditional orthogonal multiple access technology in the case of low QoS constraint.2.To solve the problem of low resource efficiency in NOMA-based F-RANs,the access modes of users in the network are dynamically optimized.There are multiple access points in this network,and the static access mode selection based on a single indicator may lead to network congestion and performance degradation.Therefore,it is of great significance to judiciously select the appropriate access modes for different users in order to improve the network performance.In this paper,two optimization algorithms are developed to solve the access mode selection problem,namely,the evolutionary game based algorithm and the reinforcement learning based algorithm,and the rate and transmission delay are balanced.The simulation results show that the proposed dynamic access mode algorithms can improve the average payoff of users by 12%,compared with the existing static access strategy based on a single indicator.The results presented in this thesis provide theoretical basis for applying NOMA technology in F-RANs as well as technical insights into the performance optimization of NOMA-based F-RANs.