Research On Access Network Performance Optimization Of 5G System

The fast development of Mobile Internet dramatically accelerates the intelligent transformation of our society. In order to establish a full-connective society, various kinds of terminals growing exponentially will gradually become part of the mobile communication system, of which M2 M devices accounts for the majority. The staggering amount of M2 M devices would inevitably impose severe side impact on the existing network system which is primarily designed for the H2 H communication. Features of huge quantity, complex services and irregular distribution greatly affect the performance of access network, which is mainly embodied on the random access procedure, resource scheduling and boundary determination among cells. It is of urgent significance to propose a series of optimization algorithms for improving access network performance under the premise of maintaining the existing H2 H devices, which is in accordance with the trend of envisioned 5G communication technologies. In order to accommodate challenges of future mobile communication scenarios, a series of improved scheduling scheme and boundary load balancing strategies are profoundly discussed in this paper, which is supposed to meet as much requirements for M2 M communication as possible while guaranteeing communication for H2 H devices. 1) Random access process in LTE system is still based on slotted ALOHA protocol, which performance would gradually reduce if the number of users is increasing. Once enormous amounts of M2 M devices compete for resources with H2 H devices at the same, the throughput of both H2 H and M2 M devices would dramatically decrease. This paper proposes a self-learning random access scheme based on virtual ID for M2 M communication, which provides services for diversified M2 M devices with different delay requirements effectively and guarantees the whole throughput of system as well. 2) Most of the M2 M devices are sensors and most of their communication requirements are uplink transmission. When the number of connected M2 M devices which are requesting services exceeds a certain amount, scheduling complexity and scheduling delay would greatly increase. This may worsely lead to much packet loss due to the limitation of resources. In order to decline the impact on H2 H users caused by massive amount of M2 M devices as much as possible, the service types are divided into real-time services and nonreal-time services in this paper. Besides, the nonreal-time services for M2 M communication apply transmission technology without scheduling based on SCMA technology, while a proposed hierarchical uplink scheduling scheme based on different serviced queues is applied to all real-time services and nonreal-time services for H2 H communication. System benefit can be maximized through altering the scheduling ratio of queues. 3) In the future 5G networks, UDN deployment would be widely employed, and as a consequent, generate even more complicated cross coverage areas caused by multiply cells. Low mobility of M2 M devices may make most of them stay in these cross coverage areas for a longer time. In this paper, a load balancing strategy among multiply cells is given based on multi-swarm particle swarm algorithm. The whole throughput of the cross coverage area is significantly optimized thought joint access control for the devices in this cross coverage area conducted by multi-cells.