Study And Implementation Of QoS Management Of Gateway In LTE Communication System

With the development of mobile communication technology, Wireless communication technology has developed from 3G to LTE. LTE technology by virtue of its high-speed, low-latency, more and more get the favour of the user. LTE technology not only greatly improved the communication speed and communication bandwidth, but also LTE communication technologies in support of business has beengreatly expanded. The user experience is not limited to the basic requirements of a normal SMS, voice, More user prefer mobile shopping, video call, VoLTE, new business which broadband wide by the mobile communication technology. Although the LTE network compared to the previous generations of communications technology in speed, bandwidth, delay have been greatly improved, but the wireless resources and network resources are still not fully satisfy all of users. One of the important means to improve the mobile communication system channel utilization and quality of service(QoS) is the rational management of resources on the network side.This thesis analyzes the overall architecture and core technology of LTE. MIMO and OFDM technology enables the LTE system can transmit data more efficiently through the use of a larger limit of limited bandwidth. But this technology will also be restricted in sharp increase the number of users, so user's request not timely appropriate in some cases. Bearer is the smallest unit in QoS, but also the implementors resource scheduling policy. Depending on the business and different needs can be configured into different priorities and types. LTE system includes eNB and GW will be carried forwarding capability by the bearer configuration. This paper from the core network point of view compares the advantages and disadvantages of resource scheduling algorithms. Although polling scheduling algorithm from the overall sense is relatively fair, but precisely because of this fair will result in its forwarding capability is limited, because the system will spend more time on user needs no forwarding. Although fair throughput algorithm balances the difference between the user throughput, but in some cases can not be done in a timely mannercorresponding burst traffic. Leaky bucket algorithm to control the exit velocity constant, but the difference can not treat different service packets, making real-time demand strong data there is a big delay. Finally token pass algorithm specified by the time slice generated fixed token data to be forwarded by the forwarding token to get the data forwarding, the algorithm is able to achieve a certain bursty traffic control,also the network can be relatively constant flow rate.In this thesis, the current LTE networks require differential treatment needs and respond to the needs of different bursty traffic control services and the user, the token bucket algorithm is proposed based on the "loans" mechanism. The method is based on the traditional token bucket algorithm allow improvements in the token shortage You can borrow the next time period to complete the current data token forwarding.Process to achieve, in order to improve the system's ability to process packets, using the Linux operating system support e_poll mechanism. The main idea of this mechanism is through the establishment of the kernel buffer to put data need to be monitored to establish the list, if the underlying data arrives when you can directly modify the cache ready list, the upper layer is required to forward data when read directly ready on the ready list of data, thus avoiding the loop through the list of the monitoring data, so you can save time, thus reducing overall system latency.Implementation section introduces several components of the system, including account data generated accounts module, registration module maintains the registry,session control of the main process module is responsible for receiving and forwarding control module receives the data transmitted from the transmitting module.Finally, the algorithm can distinguish the different treatment of business data to ensure and Between different business about 5ms time delay that the strong real-time data to give priority to forwarding delay data.