The Research Of LTE Baseband Software Processing Based On ARM

After years of development,wireless communication system has made great progress.A plenty of transmission and standards have occurred and will coexitst for a long time.Nowadays,one base station generally only support one standard,which will cause the follow-up upgrade difficult and hardware utilization low when new transmission standard appears.So a variety of transmission standards implemented by th same platform has become an urgent requirement.This is the concept of software radio.In recent decades,chip design technology and manufacturing technology are being developed by leaps and bounds.The processing capacity of general-purpose processor is increasing.In the mean time,the low-cost and high-performance RISC microprocessor ARM are widely used in industry,control,digital mobile terminals,network equipment and etc.Thus,how to implement LTE uplink baseband soft processing on ARM processor is a real problem both in theory and practice.By the 863 project “the research on soft processing with ultra-high throughput for 5G base station testbed”,we studied the LTE standards and algorithms in physical layer in detail.The transceiver for uplink shared channel is implemented and verified on ARM JUNO platform.In order to improve the processing speed,we did som dedicated optimization for time-consuming part in the transceiver.It is necessary to carry out the transmission precoding process before the SC-FDMA symbol is generated.This process involves a large number of complex multiplication operations.The direct calculation is very time consuming.Therefore,a hybrid algorithm is proposed.Besides,MAP algorithm also contains a large number of multiplication,division and exponentiation operations.Therefore,the improved LOG-MAX-MAP decoding algorithm is used to reduce the complexity,and the operation efficiency is improved by splitting the long frame into multiple subframes.And decoding in parallel.The ARM-based single-instruction multi-stream system NEON technique can effectively achieve parallel processing,so we use the NEON instruction on the transmission precoding module and turbo decoding module.