In the website from Qualcomm you can find two application examples of Raptor Q codes , which are a special type of fountain codes:

http://www.qualcomm.com/media/documents/real-image-digital-cinema-distributuion-using-raptorq

http://www.qualcomm.com/media/documents/raptor-codes-mobile-multimedia-broadcasting-case-study

Concretely, the applications are delivery of digital cinema movies over satellite to a many of cinemas and a solution for videoconferencing.

Raptor codes are also part of the MBMS standard and a proposed standard from IETF (RFC 5053, RFC 6330).

At DLR, where I am working, we have developed a working implementation of a fountain encoder/decoder, and we have demonstrated them in the lab. Currently there are plans to launch a satellite which can perform onboard fountain encoding to deliver files to several satellite terminals. For more information see:

http://www.dlr.de/kn/en/desktopdefault.aspx/tabid-4309/3222_read-26607/

Hi Chris,

There are many codes that belong to the family of Fountain codes. The most famous among them are LT codes and Raptor Codes. The main attractive feature they have is low-complexity encoding/decoding. They can also be generated randomly. They can be used for communication and storage. If you want an excellent read, I recommend an article "Fountain Codes" by MacKay D.J.

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=1561992&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D1561992

In the website from Qualcomm you can find two application examples of Raptor Q codes , which are a special type of fountain codes:

http://www.qualcomm.com/media/documents/real-image-digital-cinema-distributuion-using-raptorq

http://www.qualcomm.com/media/documents/raptor-codes-mobile-multimedia-broadcasting-case-study

Concretely, the applications are delivery of digital cinema movies over satellite to a many of cinemas and a solution for videoconferencing.

Raptor codes are also part of the MBMS standard and a proposed standard from IETF (RFC 5053, RFC 6330).

At DLR, where I am working, we have developed a working implementation of a fountain encoder/decoder, and we have demonstrated them in the lab. Currently there are plans to launch a satellite which can perform onboard fountain encoding to deliver files to several satellite terminals. For more information see:

http://www.dlr.de/kn/en/desktopdefault.aspx/tabid-4309/3222_read-26607/

Thank you gentlemen. I do have a passive appreciation for Fountain codes, but I have seldom seen them being used commercially for any particular application. I appreciate their capabilities to handle deletions and can be a resourceful way of managing high noise situations. My expectation is that they could have a place in certain satcom environments or maybe even data storage media, but it would be good to get some info on which application areas they are best suited for. Any suggestions ?

Francisco, thank you for the example applications. I was interested to consider Inter-satellite links (M2M) for file transfer in the context of EDRS type LEO to GEO communications. Could this type of mission requirements be part of your NEXT project ? You can imagine the scenario where a LEO Earth observation satellite wishes to 'dump' large amounts of data via a GEO satellite, but prefers not to have a heavy load on a return channel for handling re-transmits or require a complete reconnect and restart transmission after a to connection failure. It seems that both the Qualcomm and your NEXT project address my interests. I have worked on satellite communications solutions for some years now and would certainly be interested to follow your work.

Christopher, the use of fountain codes generally makes only sense when you have an erasure channel. In the case of an inter-satellite link from LEO to GEO using a fountain code would make sense whenever you link is interrupted. For example:

- If you use an optical link this will probably be the case if you have pointing errors. Pointing errors result in long error bursts, hence, the channel can be considered an erasure channel.

- If your link is interrupted for any reasons, this can be also seen as an erasure channel.

Another advantage of fountain codes with respect to using data transfer protocols is that you need to use less the return channel. Maybe this also plays a role in your scenario.

Fountain codes are particularly useful when the erasure rate of the channel is not a-priory known. If the erasure rate of the channel is known, it is possible to use a fixed rate erasure correcting code (LDPC, for example).

There are different types of fountain codes with different enconding / decoding complexity. If you want to do encoding onboard in the LEO I guess you will have some complexity constraints. In our NEXT project, we will do encoding onboard a GEO and this results in complexity constraints. Also the amount of memory needed will play a role. For example, Raptor encoding requires quite a lot of memory, since you need to store the intermediate symbols. Other solutions use much less memory at the transmitter. I hope my answer helps.

Hi, rateless codes are also exploited in multicasting over cellular networks, as shown for instance in these papers

Multimedia Broadcast with fountain code in Cellular CDMA

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5407203

Efficient GF(256) raptor code decoding for multimedia broadcast/multicast services and consumer terminals

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=6227434

An Improved Algorithm of 3GPP MBMS Raptor Codes

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5459886

I much appreciate you inputs gentlemen. Very useful information, for which I am sure many of your reading suggested material will help raise even more questions which I hope I can bounce of you in future correspondents.

Please see www.codornices.info

Founsure(Article Founsure 1.0: An Erasure Code Library with Efficient Repair ...

I can understand that Fountain codes can be used for broadcasting. It was mentioned that Fountain codes can also be used in data storage applications. I cannot understand how fixed-rate (they must be fixed rate as the 'receiver' is reading, i.e. 'receiving', later in time, may be even years later) Fountain codes can be of real use since Reed Solomon codes (MDS) have less redundancy for the same erasure rate decoding.

Perhaps (encoding/decoding) complexity could be one of the reasons, or parallel implementability could be another. In my humble opinion, whenever the redundancy is not the only optimisation criterion, (there are many of those criteria btw: network bandwidth, locality, complexity, hardware-adaptability, reliability, availability, security etc) different code structures may offer different points in the overall trade-off curve. Particularly, based on my experience, a new set of critieria almost always emerges in the standardisation processes and application development phases...Unique features of fountain idea (both linear and non-linear versions) could help in some if not all of these cases.

Sure, I fully agree with you, code redundancy is not the only design parameter of a given system, and there are many others such as the ones you mention. In a standardization process, however, code redundancy is of prime concern, and it will never get better over the, say, 10-15 years lifetime of a standardized product, that is, the redundancy loss with respect to the alternative, less redundant, method can never change. The relevance, however, of the other parameters, such as hardware complexity etc, diminish over time with Moore's law. In other words, I suggest, one must weigh the hardware issue by looking ahead, say, ten years.

I see. Your observation I believe is true specifically for complexity perspective. Perhaps, that's why Qualcomm's RaptorQ was designed to minimise coding overhead (one or two extra symbols for perfect recovery). However, considering other alternatives, sometimes bandwidth becomes extremely important in a talkative network. This happens in IoT world pretty frequently as well as platforms that are based on gossip-type protocols. Now thinking about coded data spread across such distributed/decentralized schemes, maybe complexity concerns are brought to daylight a bit later I presume say in an architecture design. Plus, having smallest possible coding overhead makes it mathematically impossible to obtain best bandwidth performance. All in all, I think it would still take tremendous industry and engineering effort to help such code structures find important use cases in our daily lives.

Hope this helps.

Article Maximization of Underwater Sensor Networks Lifetime via Fountain Codes