基于GPP平台的LTE上行信道估计高效实现

处理的实时性是基于GPP(general purpose processor)平台的LTE基站系统的关键指标。在上行信道估计环节,往往需要维纳插值来获得良好性能;但其过高的计算复杂度使信道估计成为上行物理信道实时性处理的计算瓶颈,需要结合GPP特点进行算法实现的优化。通过引入部分线性插值和预存维纳插值系数的方式,显著降低维纳插值的运算复杂度。再使用SSE指令集和"流水线"处理的优化,在保持性能基本不变的情况下,在单核处理器上处理100个物理资源块的上行信道估计仅需6.36μs,为优化前的1.5%,使得高性能上行信道估计不再成为GPP软件无线电实时处理的瓶颈。

Efficient Implementation of LTE Uplink Channel Estimation on GPP Platforms

In recent years, as the rapid development of mobile data services, 3GPP Long Term Evolution has become a significant communication system. At the same time, China Mobile has proposed the C-RAN project, which aims at the evolution of green radio access network. C-RAN is based on general purpose processors, leveraging the software defined radio technologies to reduce the energy consumption and realize the stable upgrade of different protocols. This paper takes the module of channel estimation in LTE uplink as an example to illustrate the procedure of using SSE instructions to accelerate software. By partially linear interpolation, pre-storing the coefficients of Wiener interpolation and pipelining techniques, we reduce the computation complexity of the Wiener interpolation channel estimation algorithm. Maintaining the estimation performance, the optimized running time is reduced to 1.5% of the original method, which satisfies the real time requirements of LTE systems.