Fuzzy Q-Learning-Based Hybrid ARQ for High Speed Downlink Packet Access

In this paper, a fuzzy Q-learning-based hybrid automatic repeat request (FQL-HARQ) scheme is proposed to enhance the system performance of high speed downlink packet access (HSDPA) in universal mobile telecommunications system (UMTS). In the HSDPA system, it is an important issue to choose modulation and coding scheme (MCS) for new data packet when channel quality indicator (CQI) is inaccurate. Hence, the purpose of the FQL-HARQ scheme is to determine a suitable MCS for new data packet so as to maximize the system throughput while guarantee the block error rate (BLER) requirement. In the FQL-HARQ scheme, the fuzzy logic is used to choose a suitable MCS for each new data packet. According to the feedback information from the HSDPA system, the Q-learning algorithm is adopted to update fuzzy rule base. This makes the FQL-HARQ scheme can adapt to the variation of environment. Simulation results show that the proposed scheme can increase the system throughput of up to 70% compared to the conventional adaptive method in the case with CQI report delay.