A GPP-DSP based computing task allocation scheme for on-board signal processing in OFDM satellite system

Abstract

Satellite communication plays an important role in the global connection for its large coverage and stability against the terrestrial disaster. Usually the on-board signal processor is customized for the satellite which it is implemented on, with high processing ability but also with high construction cost and difficulty of reconfiguration. Thus a reconfigurable on-board processor is required to fulfill the demand of more frequent and periodic configuration update. This paper presents a computing task allocation scheme based on general purpose processor, (GPP)-digital signal processor (DSP) signal processing system, in which GPP handles the generic algorithms such as channel coding, modulation, time and frequency synchronization while DSP is in charge of special mathematic operations. We first introduce a low earth orbit (LEO) orthogonal frequency division multiplexing (OFDM) satellite physical layer system as an example with an analysis of the system's calculation characteristics. After that, a computing task allocation scheme corresponding to the system is derived based on the computation characteristic of all the modules in the system. Finally, the processing time with and without the scheme algorithm is tested on a system consist of a PC and a DSP board. Experiment results show that the processing time of the DSP-GPP system is shortened using the proposed scheme, compared to the GPP-only system without the task allocation scheme.