面向软件无线电的物理下行共享信道优化与实现

针对软件无线电平台Open Air Interface(OAI)终端侧物理下行共享信道(Physical Downlink Shared Channel, PDSCH)中解扰模块和信道译码模块数据延时大的问题，提出了基于图形处理单元(Graphic Processing Unit, GPU)中统一计算设备架构的PDSCH优化方案，对传输块进行比特级和码块级数据划分，设计了多GPU下的并行解扰和并行信道译码，降低了数据延时，提高了下行峰值速率。实验结果表明，该优化方案基本维持了原有循环冗余校验码错误率，且在不同传输块大小下，解扰模块耗时最大降低92.1%,信道译码模块耗时最大降低83.7%,终端PDSCH耗时最大降低77.8%,下行峰值速率最大提高138.3%,有效提升了OAI平台性能。

Optimization and Implementation of Physical Downlink Shared Channel Oriented to Software Defined Radio

For the problem of large data delay in the unscrambling module and channel decoding module in the physical downlink shared channel(PDSCH) on the terminal side of the software defined radio(SDR) platform Open Air Interface(OAI),a PDSCH optimization scheme based on the compute unified device architecture(CUDA) in graphic processing unit(GPU) is proposed.In this scheme,the transport block is divided into bit-level and block-level data,parallel unscrambling and parallel channel decoding under multi-GPUs are designed,which reduce the data delay and increase the downlink peak rate.The experimental results show that the optimization scheme basically maintains the original cyclic redundancy check(CRC) error rate,and under different transport block size,the time consumption of the unscrambling module is reduced by 92.1% at the maximum,the time consumption of the channel decoding module is reduced by 83.7% at the maximum,the time consumption of the terminal PDSCH is reduced by 77.8% at the maximum,and the downlink peak rate is increased by 138.3% at the maximum,effectively improving the performance of the OAI platform.