Research On Adaptive Modulation And Coding Technology In Td-Hsdpa

In order to meet the growing demand for data services, HSDPA was introduced in3GPP’s Release-5specifications to improve the downlink throughput. AMC is one of the key techniques in HSDPA which periodically changes the modulation and Coding Scheme according to the varying channel condition. And in the AMC procedure of TD-HSDPA, there are three sources which can affect the performance of AMC: channel quality estimation, the selection of MCS based on SNR, and feedback delays. In this paper, we mainly focus on the last two problems.First of all, to solve the second problem,"The fixed MCS threshold based on SNR" is widely used. In this method, we should simulate the performance of all the MCS to get the thresholds. However, in TD-HSDPA system, it defines64kinds of transmission block size and3kinds of modulation for each UE and makes the combination of transmission block size and modulation complicated. As a result, a lot of simulation should be done to get the switching threshold of different MCS. Therefore, a new SNR to CQI mapping method which can reduce the amount of simulations is proposed in this paper. What’s more, since the switching thresholds are obtained via the simulation under a specific scenario, they are not the optimal ones in other scenarios. So some threshold adjustment algorithms are proposed. The traditional centralized threshold adjustment and decentralized threshold adjustment method cannot put attention to convergence speed and fluctuation of the thresholds simultaneously. So an improved threshold adjustment algorithm which can solve this problem is put forward in this paper.Secondly, to compensate the loss of performance caused by feedback delays, some channel prediction algorithms are proposed, such as linear predictor based on NLMS (Normalized Least Mean Squar), Kalman filter, predictor based on AR model and so on. However, the complexity of these algorithms is very high, and their performances in high speed scenario are not very good. Therefore, a new CQI prediction algorithm which works well in high speed scenario and with low complexity is proposed in this paper.