Research On Resource Allocation Algorithms In 5G Radio Access Networks

The Fifth Generation as a new generation mobile communication system for mobile communication development after 2020,it must have high-speed,low-latency,wide-connection and other characteristics to meet future mobile communication demands.In order to achieve the above goals,many novel mobile communication technologies are constantly emerging.For example,Non-Orthogonal Multiple Access technology utilizes power domain multiplexing to make up for low spectral efficiency of traditional Orthogonal Multiple Access technology.In addition,network architectures of communication system are constantly evolving.For example,Cloud Radio Access Network is gradually developing towards the trend of Heterogeneous Cloud Radio Access Network.Although network resource allocation has existed for a long time,it still plays an important role in5 G development,and it also needs to adapt to the new demands of 5G development.Therefore,this thesis deeply studies 5G network resource allocation with the aim of enabling network resource allocation to be combined with novel mobile communication technologies and network architectures,and to adapt to the new demands of 5G development.The main researches of this thesis are as follows:(1)A MIMO-NOMA network resource allocation algorithm based on user clustering is proposed.The algorithm clusters users based on their location and Channel State Information(CSI),which effectively mitigates co-channel interference caused by sharing the same resource block.At the same time,the algorithm introduces effective capacity as a measure of network performance.Compared with traditional network capacity,it can guarantee various Quality of Service(Qo S)requirements.The simulation results show that user clustering can alleviate the co-channel interference introduced by NOMA technology and improve the overall network performance.At the same time,effective capacity also guarantees various Qo S requirements such as rate,delay and packet loss rate.(2)A H-CRAN resource allocation algorithm based on multi-time scales collaborative optimization is proposed.The traditional resource allocation algorithms allocate resources on small-time scale based on short-term instantaneous information such as channel gain.The algorithm proposed in this thesis allocates resources on large-time scale based on long-term average information such as historical bandwidth requirements before small-scale resource allocation.At the same time,the algorithm combines machine learning methods to solve problems,and can betteradapt to the dynamically changing network environment than traditional optimization methods.The simulation results show that multi-time scales collaborative optimization can further improve resource utilization compared to existing resource allocation algorithms.(3)A demonstration platform of H-CRAN resource allocation system is designed and implemented.The platform is built on the utility of Jet Brains Py Charm and relies on multi-threading technology to implement distributed computing.It mainly uses library functions such as Numpy and Pandas to analyze and process network data,and uses framework such as Tenseflow and Keras to implement network resource allocation.The platform implements H-CRAN resource allocation algorithm based on multi-time scale collaborative optimization and verifies its performance.