Research On Satellite Communication Based On Fountain Code

Satellite communication is playing an important role in global communication, however, there are still some challenges facing by satellite communication, the most notorious among which are path loss and propagation delay. Traditional communication mechanisms, such as FEC and ARQ, can’t satisfy the demand of high reliability and low delay at the same time. Fountain code is a kind of technology which was first applied in multimedia broadcast. The basic idea behind fountain code is avoiding feedback, which may reduc the delay indirectly. Besides, fountain code always employs powerfull error detecting technique (such as CRC) to assure that only correct packets are accepted. So, fountain code can take care of both reliability and delay. In this sense, fountain code is really promising in satellite communication. However, the application of fountain code in satellite communication still faces some challenges, that’s where our research and contribution lie on.1) The first challenge is that traditional decoding algorithm, such as BP algorithm and Inactivation algorithm, will waste lots of redundancy, because they only dispose the correct packets. Some researchers employ Log-BP algorithm to ultilize the information conveyed by the corrupted packets, however, the computation cost of Log-BP algorithm is usually unacceptable. To deal with problem of waste, we propose a class of novel cross-level (CL) decoding algorithm for fountain code. In application layer, CL algorithm employs traditional decoding algorithms, such as BP algorithm and Inactivation algorithm. In physical layer, CL algorithm introduces a novel decoding procedure, called Mining Decoding, to deal with the corrupted packets. In Mining Decoding, we turn to the existing FEC (such as LDPC), but not Log-BP algorithm, to do some recovery from the corrupted packets, which makes it very compatible to the existing system. The most impressive advantage of the proposed CL algorithm is that it imporves the decoding performance significantly on the cost of limited extra computation. It helps to mitigate the waste of redundancy, and as a result, lower the acception delay.2) The second challenge is that fountain code always needs redundancy to complete the decoding, even if the channel is good enough, on the other hand, if the channel is really bad, the redundancy of fountain is not so efficient as to that of HARQ. So, in some scenarios, fountain code can’t always keep lower delay. To deal with this problem, fisrt we analyze why and when SR-HARQ (mainstream communication mechanism based on feedback) incurs serious acception delay, and when fountain code can achieve lower delay. Specifically, we propose a nove metric called relative delay to describe the waste of transmitting capacity caused by feedback. Higher relative delay always means more transmitting capacity is wasted, in which sense, means fountain may perform better in term of accpetion delay. In our paper, we will perfom lots of end-to-end simulation to claim the advange area of fountain over SR-HARQ in the context of kinds of orbit altitude, data rate, data volume, channel quality and decoding ability. Our work will be a powerful guideline for the protocol designer, who may need to make a trade off between fountain code and SR-HARQ.The work in our paper is very practical. Firstly, it improves the decoding robustness of fountain code in wireless communication. Secondly, it promotes the application of fountain code in satellite communication, making us have more choice when we need to make a tradeoff between reliability and delay.