Chapter 1006 Maximum Rank Distance Code

In fact, Zhang Weiyong’s earliest professional background was in communications technology.

However, after he started serving as chief satellite engineer in the mid-to-late 1990s, his main focus was on the overall engineering aspect.

Of course, even if he was still specialized in his original major, he might not have noticed the paper that proposed network coding in advance——

At the beginning of the 21st century, we are catching up with the explosion of network communication technology. Not to mention every day, at least every month new concepts and new technologies are proposed, and almost every proposer claims that their research can change the world.

Even a research team with abundant resources cannot pay enough attention to every result.

Not to mention that China’s aerospace research has always been focused on stability and is relatively conservative in the application of new technologies.

But if it’s a technology that’s been proven in other industries… that’s another story.

Therefore, after hearing Chang Haonan's answer just now, Zhang Weiyong immediately sat up straight:

"Mr. Chang still has research on information theory?"

Data transmission and data compression are both categories of information theory.

Chang Haonan has not had any research results in this area before.

But after thinking about it, it seems that Chang Haonan has expanded into a new research field, which is not very surprising...

"Although network coding does belong to information theory, I came into contact with this field purely from a mathematical perspective..."

Chang Haonan explained with a smile:

"You know, I have some research on topology and topological structures..."

"Cough cough cough——"

At this moment, Zhang Weiyong just picked up the cup again and took a sip of wine, but was choked by Chang Haonan's words.

A man who solved the Poincaré conjecture and created a manifold learning algorithm.

Say you have "some research" on topology?

That's really well researched.

After a while, Zhang Weiyong finally calmed down and waved his hand to express that he was fine, and motioned Chang Haonan to continue performing... uh... keep talking.

"In short...some time ago, someone came to me to review a paper on nonlinear coding algorithms. I found that there were several very basic theoretical issues in his paper. It was definitely impossible to pass the paper, but the ideas he proposed It’s really new, so I stopped by to learn about the research background. ”

After Chang Haonan drank a small glass of red wine, he did not continue. Instead, he got a thermos from somewhere and drank water instead.

While talking, he also poured a glass for Zhang Weiyong.

"So you're going to study it in depth?"

"Research is out of the question."

Chang Haonan shook his head:

"Things in the field of information theory are okay at the theoretical level. If you really want to study engineering applications, you will need enough time and funds. At most, I have a little understanding..."

At this moment, Zhang Weiyong finally understood.

The other party went around in such a big circle because he originally wanted to apply for the project with him, and then take advantage of the resources at the Aerospace Port.

After all, not to mention that Chang Haonan is only a half-step academician now. Even if he is really a big brother in the Dzogchen realm of academicians, it is impossible for the country to build a batch of satellites or build an entire dedicated network just by talking about it. Carry out technical verification.

It must rely on some names.

Of course, Zhang Weiyong would not believe such nonsense as "I just learned about it and haven't studied it in depth yet".

"Can you tell me more?"

He asked calmly:

"I mean, about that specific broad code."

"certainly."

Chang Haonan turned the piece of paper to the back:

"We first simplify a multicast directed acyclic network with single source and dual sinks. All point-to-point links in the network have unit capacity. S is the source, T1 and T2 are sinks, and the rest are intermediate nodes... …”

As he spoke, he drew two schematic diagrams on the paper.

"According to the 'maximum flow minimum cut' theorem, in any directed graph, the maximum flow of information from the input node to the output node is equal to the capacity of the minimum cut set that separates the input node from the output node. Then the multicast network in the figure below The maximum theoretical transmission capacity is 2 bits/unit time..."

"If we follow the traditional model in the picture on the left, since link CD is the bottleneck of the network transmission, it can only transmit 1 bit of information. As a result, node C must use the traditional queuing storage method to realize information transmission, resulting in the throughput of each sink node. The amount is only 1.5 bits/unit time..."

"In the picture on the right, node C encodes the received message a and message b... For simplicity, we choose the simplest XOR logic. In short, the sink node can use the received message a or b, as well as a and b The encoded result is simultaneously translated into another unreceived message bit, thus reducing the transmission of one information stream and increasing the throughput of each sink node to the theoretical upper limit of 2 bits/unit time..."

"Of course, due to the existence of the Shannon limit, it is impossible for any technology to compress data infinitely, but this is just the simplest case. You and I both know that the more complex the information transmission network is, the difference between the actual throughput and the theoretical transmission capacity will be. The wider the gap, the more obvious the potential of this type of algorithm..."

"..."

"Wait a moment……"

Just when Chang Haonan was about to reach the end of his talk and the paper was finally almost full, Zhang Weiyong suddenly stretched out his hand:

"Mr. Chang, what you just said has a big premise, that is, it is assumed that there are no errors in the transmission process... If any of the nodes on the right has a coding error or a transmission error, then the nodes behind will be infected, causing the error to spread to the entire communication node like a computer virus..."

"Not to mention that such errors are very common in normal communications, military satellites also have to consider the opponent's electronic interference..."

The latter's concerns are naturally justified.

The traditional information relay method repeatedly sends data, although it will cause jams and delays, but it can still send the correct information.

But if something goes wrong in the middle of the encoding and decoding process of this new trick that Chang Haonan is going to do, then the relevant data may be wrong.

Moreover, due to the extremely fast spread of errors, it is still difficult to figure out which step the problem is.

In the end, the data has to be sent repeatedly, which will only become more jammed.

However, facing this question, Chang Haonan put on a matter-of-fact look:

"That's why I said that it still requires in-depth research and technical verification..."

"If I can come up with the ready-made technology now, why would I need to find you? Can't I just go to Director Luan directly..."

"Hiss..."

Zhang Weiyong scratched his head and felt that it seemed to make sense...

"In short, Mr. Zhang, you have just mentioned that the focus of this technology now is to develop a coding algorithm with stronger compression capabilities, and secondly, to find a way to find an error correction mechanism for the information transmission process, so that even if the data goes wrong during transmission, it can be corrected at the next node to ensure that the error will not spread between different nodes."

Chang Haonan said, reaching out and tapping the paper lightly to emphasize.

At this point, Zhang Weiyong entered a relatively familiar field:

"For example... maximum rank distance code?"

"MRD code can indeed work."

Chang Haonan was a little surprised that the other party could directly point out the answer he wanted, but his expression did not change much:

"However, the current research on MRD is still mainly focused on using it for data encryption, and the error correction function is just incidental... So if you want to achieve strong error correction capabilities, you must develop new MRD code technology."

"I can use the resources of Torch Group to conduct some research and testing in a small-scale local area network and aviation relay communication field, but these environments are still too mild compared to satellite communications. It will definitely require a... or even a group of special satellites to conduct intersatellite link testing to finally determine the feasibility of using it in the aerospace field..."