Rebirth: Starting from Lighting Up the Tech Tree

Chapter 30 Undercurrents

The work on solar storms has been much smoother than that on torrential rains.

The reason is simple—the impact of solar storms on the space-to-ground link is mainly transmitted through the ionosphere, and Lin Ke had already taught him how to handle ionospheric refraction. Zuo Cheng added an ionospheric disturbance early warning module to the bottom layer of the two-layer architecture, which connects to real-time data from the Blue Star Space Weather Monitoring Center. When a solar storm causes severe ionospheric disturbances, the module will automatically increase the compensation intensity.

Completed in three days. Simulation accuracy: 33% above the baseline.

Next came multi-satellite switching, low elevation angle fading, multipath scattering, polar coverage, and ocean propagation. Zuo Cheng tackled each of these five scenarios in twelve days.

Each scenario has its own challenges, but Zuo Cheng has figured out the pattern—the two-layer architecture is robust enough that most problems only require adding corresponding physical correction terms to the lower layer, and the adaptive compensation in the upper layer can cover the remaining errors. There are very few situations where the architecture actually needs to be modified; it's more of a process of "filling the framework with ammunition."

On the thirty-first day, all seven extreme scenarios were successfully completed.

The worst scenario—polar coverage—had an extremely long signal path due to satellite tilt limitations, resulting in accuracy "only" exceeding the baseline by 28%. The other six scenarios all exceeded 30%.

Zuo Cheng compiled all the simulation results into a summary report and drew seven checkmarks next to the list in his notebook.

There are eleven days left until the joint review. The technical framework of the solution is complete; the remaining work involves refining the details, supplementing documentation, and preparing presentation materials.

But at that moment, something unexpected happened.

On the morning of the thirty-second day, Zuo Cheng walked into the twenty-third floor work area and found that the atmosphere was not right.

Lin Ke sat at her workstation, her face ashen. Tang Xu stood beside her, his lips pressed tightly together, saying nothing. Cheng Yuan was not there; his computer was on at his workstation, the screen displaying the standby image.

"What's wrong?" Zuo Cheng put down his backpack.

Lin Ke looked up and handed him a printed document.

"Take a look at this."

Zuo Cheng took it and glanced at it.

It was a summary of a technical solution titled "A Deep Learning-Based Channel Prediction Scheme for Satellite-Ground Links." The author's name was Cheng Yuan.

Zuo Cheng quickly skimmed through the summary, and when he got to the third paragraph, his brows furrowed.

Cheng Yuan's solution centers on using deep neural networks to make end-to-end predictions of channel changes in satellite-to-ground links—inputting satellite orbital parameters and historical channel data, and outputting the channel state at future moments. The idea itself is not new; there are many papers in academia that have done similar work.

But a sentence in the third paragraph of the plan caught Zuo Cheng's attention: "Borrowing from the idea of ​​hierarchical prediction, the deterministic factors of the orbit and the random factors of the channel are processed separately, modeled separately, and then fused for prediction."

Stratified prediction. Separation of deterministic and random factors.

This is the core idea of ​​Zuocheng's two-layer prediction architecture.

"When did he write this proposal?" Zuo Cheng asked.

"It was submitted last night." Lin Ke's voice was low, but her anger was palpable. "This morning I saw a new document in the project team's shared folder, and as soon as I opened it, I realized something was wrong. His 'layered prediction idea' is exactly the same as your two-layer architecture—except he replaced your adaptive tracking with a deep learning model."

Zuo Cheng read the document from beginning to end.

Cheng Yuan's solution is indeed fundamentally different from his—the technical approaches are completely different; one uses adaptive signal processing, and the other uses deep learning. However, the core idea at the framework level—separating deterministic and stochastic factors for processing—is Zuo Cheng's original insight, and no publicly available literature had proposed this approach before him.

Cheng Yuan used this idea in the plan, but did not cite the source or communicate with Zuo Cheng in advance.

"He probably thinks this is just a general methodology, nothing particularly original," Tang Xu said, his tone calmer than Lin Ke's. "But the problem is, during the time he worked on the project team, your two-tier architecture solution was always on the whiteboard in the workspace; there's no way he didn't see it."

Zuo Cheng put the document back on the table and remained silent for a few seconds.

This incident reminded him of Ma Hao.

The same old trick—not a direct copy, but "borrowing." A makeover, a different technical approach, but the core insights are someone else's. Strictly speaking, it's not plagiarism, but the way it's done is unsightly.

However, there is a crucial difference between Cheng Yuan and Ma Hao—Ma Hao openly presented Zuo Cheng's solution as his own in the group meeting, while Cheng Yuan merely cited an unattributed idea in his independent solution. The former was stealing credit, while the latter was skirting the rules.

"What are you planning to do?" Lin Ke asked, looking at him.

Zuo Cheng thought for a minute.

"I don't know what to do."

Lin Ke was stunned.

"During the review process, the judges look at the quality of the entire solution, not who came up with a particular idea first," Zuo Cheng said calmly. "My solution forms a complete closed loop, from theoretical derivation to simulation verification to engineering design, with solid data support at every stage. Cheng Yuan is using deep learning in this approach, which requires a large amount of training data and computing power. Within the six-week time limit, it's difficult for him to achieve the same level of verification depth as me."

He glanced at Lin Ke.

"My victory wasn't based on 'who came up with the idea first,' but on 'who implemented the idea better.' If I went to complain to President Zhou that Cheng Yuan had borrowed my idea, even if President Zhou agreed with my statement, it would only make me look petty in front of the judges."

Lin Ke opened her mouth, seemingly wanting to refute, but in the end she didn't say anything.

Tang Xu nodded in agreement: "You're right. Letting your work speak for itself is more important than anything else."

In the afternoon, Cheng Yuan returned to the work area.

His demeanor was no different from usual—his gold-rimmed glasses were clean, his shirt was buttoned up to the top, and he sat down, turned on his computer, and started typing code.

Zuo Cheng noticed that Cheng Yuan glanced at the printed document on his desk as he entered—the same summary Lin Ke had shown him. Cheng Yuan's gaze lingered on the document for less than half a second before casually looking away.

He knew Zuo Cheng had seen it.

Zuo Cheng also knew that he knew.

But neither of them spoke.

Over the next few days, the atmosphere in the work area became subtle. Lin Ke's tone turned noticeably colder when she spoke to Cheng Yuan, while Tang Xu maintained a polite facade but stopped discussing technical details with Cheng Yuan. Cheng Yuan seemed completely unconcerned, arriving and leaving on time every day, coding rapidly as if nothing had happened.

Zuocheng maintained absolute calm amidst these undercurrents.

He devoted the last eleven days to refining the solution—running each simulation scenario twice to confirm reproducibility, polishing the technical documents word by word until there was no ambiguity, preparing three versions of the defense materials, and preparing solutions for every type of question the judges might ask.

On the night of the thirty-eighth day, Zuo Cheng finished the last round of checks in his dormitory and closed his laptop.

Tang Xu looked up from the opposite bed: "All done?"

"nailed it."

"So, during tomorrow's mock defense, could you listen to my antenna design and check for any flaws?"

"no problem."

Tang Xu turned over, preparing to sleep, then suddenly said, "Zuo Cheng, you have one redeeming quality—you don't get angry when you should, and you don't hesitate to act when necessary. I used to think that kind of person was cunning, but now I think it's not cunning, it's foresight."

Zuo Cheng smiled but didn't reply.

After turning off the lights, he lay in the darkness, staring at the ceiling.

The joint review is in four days.

His plan has been prepared to the limit of what is possible.

All that's left is to go on stage.

A line of text quietly appeared on the screen in my consciousness:

[Main Quest Chain - Breaking the Communication Barrier - Stage Four Progress: 89%]