Research On Protocol And Auto-Scaling For Micro-Service In 5G Core Network

With the rapid growth of mobile traffic and the unceasing increasement of use cases,the 5G network will be applied to more vertical industry.Updating legacy network and using cloud-native service-based architecture(SBA),become the effective approach for the operator to deal with diverse service and communication scenarios.However,SBA and the internet of everything become a challenge for the elasticity and flexibility of the 5G network not only,but also make a request for the resource utilization.Expanding open and improving resource utilization,are meaningful for the 5G network to increase the value of the link and reduce the operating expense of the operator.Firstly,this paper designed a complete protocol stack for the service-based interface(SBI)of SBA.Meanwhile,this paper analyzed the characteristics and made the performance evaluation for the candidate protocol.The result shows the proposed protocol stack not only meets the current requirement of the 5G core network but also is suitable for the long term evolution.Secondly,to solve the autoscaling policy of the service-based 5G core network,this paper design a self-adaptive and self-learning autoscaling framework based on reinforcement learning,which can be loaded with a priori knowledge.By taking the fuzzy Q-learning algorithm,it achieves online learning and can improve resource utilization with the performance guarantee.