A DSP-GPP Based Computing Resource Allocation Scheme For On-board Signal Processing

Satellite communication plays an important part in the global connection.Researchers have been devoting themselves into developing new communication technologies and schemes with higher efficiency and more stability for satellite communication.However,the current on-board processors are specially tailored for the target systems,which can efficiently operate the functions and techniques they are designed to,resulting in expensive costs and difficulties for new technology deployment and setup.Thus a new on-board processor structure is needed,which can meet the demand of rapid technological replacement.Under the above circumstance,this paper focus on the research of reconfigurable on-board signal processor and relevant technologies.The computational characteristics and resource needs of on-board signal processing modules of OFDM satellite,including channel coding module,constellation mapping module,IFFT/FFT module and time&frequency synchronization module,are analyzed in this paper.Resource allocation scheme of each module and the whole system is derived after the computational need analyze of each module.And the processing speed of the DSP is furtherly improved by optimizing the code with DSPlib functions and changing the storage position of the code section to reduce the CPU cycles each read/write operation will consume.At last,the proposed scheme is tested on the TMS320C6748 DSP and the regular PC as a demo system,the results shows that the scheme this paper proposed managed to use DSP to cooperate with the PC and accelerate the processing speed of the whole system.Moreover,reconfigurability has been introduced into the structure,while ensuring the performance of the whole system.Thus the feasibility of the scheme is proved.