| With the development of LTE wireless network technology,high transmission rate,low transmission delay and more stable transmission effect are required for the network.As an important part of software defined radio(SDR),digital signal processor(DSP)plays an important role in the realization of LTE wireless network system.In order to improve the performance of the system and bring users a better network,it is necessary to optimize the process of DSP.Therefore,this paper will take LTE wireless network system as the background,investigate the optimization methods of TMS320C6678 multi-core DSP,and design specific optimization schemes for different functional modules in order to maximize the performance of the system.The main work of this paper is summarized as follows:The software optimization methods based on DSP are investigated,including compiler optimization,C language optimization and assembly optimization.The modulation module of LTE transmitter is taken as an example.Firstly,this paper analyses the problems of this module,then designs a detail software optimization scheme of C language,using compile control instructions,inline functions and keywords.Finally tests the efficiency of optimized module.The results show the module runs more efficient than before.The characteristics of demodulation algorithm in LTE receiver are investigated.Based on the factor of the demodulation module has high complexity of data processing,this paper designs the C language and assembly optimization schemes.Then introduces the flow of assembly optimization scheme and lists the advantages and disadvantages of this scheme.Finally,two schemes are tested respectively.The results show the module optimized by assembly is obviously better than C language.The architecture of multi-core DSP chip and the technology of multi-core programming are investigated.A multi-core parallel data processing scheme based on interrupt processing is designed.Then the scheme is applied to the demodulation module of this system.Based on the demodulation module,multi-core task and data partition are designed.Finally,multi-core parallel experiments are carried out to verify the effect of multi-core parallel acceleration.The results show that using multi-core parallel processing can further improve the efficiency of the program. |
