Queuing Theory Based Error Control Mechanism Modeling And System Performance Analysis In LTE Network

Co-development of mobile communication technology and intelligent terminals, not only provides a good transmission channel and application platform for mobile services, but also promotes more and more people start using mobile services via mobile phone. Therefore, it requires a communication system for not only a high transmission rate, but also a certain protection of user experiences. In the LTE system, the mobile services are divided into two categories of delay-sensitive, errors-tolerate real-time services and error-sensitive, delay-tolerance non-real-time services. Each service is identified by QoS Class Identifier (QCI). For real-time services, system uses radio link control layer (RLC) automatic retransmission request (ARQ) and media access control (MAC) layer hybrid automatic retransmission (HARQ) double error control mechanism to ensure data transmission. For error-sensitive, delay-tolerance non-real-time services, system uses single error control mechanism. In order to improve the system throughput based on the basic QoS requirements, LTE system uses the adaptive modulation and coding schemes (AMC) in the physical layer, dynamically select modulation and coding schemes (MCS) based on the channel conditions, so as to find the best compromise between throughput and BER. This article relies on cooperation project "Fusion of mobile services and communications network under the4G network background" with Huawei Technologies Co., Ltd. and launched a deep study on LTE system protocol stack of access layer, error control mechanism, and the AMC mechanism. We summarized the data process in each protocol layer and laid a theoretical foundation for modeling system. To reduce system overhead and implementation complexity, LTE distinguish different RLC layer transmission modes for different services, the rest of other layers remain fixed processing modes. This mechanism didn’t take full considerations in service property, and it will reduce the system resource utilization and affect the user experience at the same time. Aiming at this problem, this paper models the important process based on Markov theory, queuing theory, and probability theory. Based on business characteristics, we introduced service differentiation in the processing model of access layer in the protocol stack, exploring the optimization problem of LTE protocol stack in theory.First, a LTE system double error control mechanism modeled by the Two-dimension Markov chain is put forward. The first dimension of the system state is the RLC layer retransmission times, and the second dimension of the system state is the MAC layer retransmission times. Next we proposed a MCS selection algorithm according to the instantaneous error rate under different MCS based on the constraints of packet loss rate. Then, we calculate different MCS choice probability and the average bit error rate based on the AMC threshold so as to make a foundation for system modeling and analysis of error control mechanism. We map the average error rate mechanism under the AMC to the state transition probabilities, and solve the average number of retransmissions under loss rate constraints. Then we get the system spectrum resource utilization. This paper gives expression of the average delay under LTE system double error control mechanism by the method of mathematical analysis. Simulation results show that the system can effectively improve the spectrum utilization by dynamically adjusting retransmissions according to the channel information. Last, this section compares the impact on system performance by the MAC layer using Type-Ⅰ and Type-Ⅱ HARQ scheme, the analysis showed that selecting Type-Ⅰ type HARQ scheme can simplify HARQ implementation while meet the system performance when the channel noise ratio is low.In order to fit the actual LTE system, this paper models the LTE RLC layer finite system cache, MAC layer HARQ scheme and physical layer AMC based on queuing theory and put forward three-dimensional Markov model. The first dimension represents the channel state, the second dimension represents the queue state, and the third dimension represents the number of retransmissions in MAC layer. Queuing system model follows Poisson distribution, servicing is mapped to different physical layer encapsulation MCS number of packets, queue length is mapped to the RLC layer cache size. The model of channel is Rayleigh fading channel. We assume that there is a limited number of MCS in the physical layer. In different MCS, each frame can encapsulation different data number, so the queuing system has multiple service rate and varies according to the channel state. By the conclusion of the previous AMC modeling, this paper map average bit error rate under different MCS to state transition probabilities and solve queuing system throughput, packet loss rate and delay by geometric matrix solution. Finally, we simulate curves of system performance under different arrival rate and compare the effects of two types of MAC layer HARQ scheme on system performance. The result shows that, for different business loss rate, there is a number of retransmissions and cache size that can make the system performance optimal.