Rebirth: Starting from Lighting Up the Tech Tree

Chapter 28 Prediction

Zuo Cheng spent the first three days doing one thing—going through all the measured channel data of the satellite-to-ground link from beginning to end.

He didn't just skim through the data; he analyzed it line by line. For each set of data, he labeled the timestamp, satellite elevation angle, weather conditions, channel gain variation curve, and Doppler shift. Over three days, his notebook was filled with more than forty pages of dense writing, and his desktop was covered with hand-drawn diagrams of channel state changes.

The other three interns were all busy with their own things. Cheng Yuan was building the simulation framework for the signal processing link, his typing so even it sounded like a metronome. Lin Ke was discussing the parameter settings of the space channel model with the chief engineer; the two argued in front of the whiteboard all afternoon. When Zuo Cheng passed by, he heard Lin Ke slam her hand on the table and say, "These parameters are physically inconsistent." The chief engineer, far from being angry, laughed instead. Tang Xu was the quietest, sitting alone in a corner drawing the radiation pattern of the antenna array. He would occasionally get up to pour himself a glass of water, and when he passed Zuo Cheng's workstation, he would glance at the notes on his desk, but never initiate a conversation.

On the third night, Zuo Cheng spread out his notebook in the intern dormitory and began to organize his thoughts.

Lanwan Communications provides decent accommodations—apartments near the R&D center, two people per room. Zuo Cheng's roommate is Tang Xu, and their relationship is simple: each does their own thing, doesn't bother the other, and occasionally exchanges a comment about the cafeteria food.

Zuo Cheng stared at the channel change curves plastered on the wall, repeatedly processing the findings of the past three days.

The channel variation of satellite-to-ground links has a characteristic that terrestrial links do not possess—it has extremely strong structural characteristics.

The channel variations in terrestrial 5G are largely random—a pedestrian walking by, a vehicle driving by, or a door opening or closing can all cause unpredictable fluctuations in the channel. However, satellite-to-ground links are different. The primary driving force behind channel variations is the orbital motion of the satellite, which is entirely calculable. The exact time and position of the satellite, its elevation angle, and the magnitude of its Doppler shift can all be precisely calculated using orbital mechanics.

In other words, the "skeleton" of the satellite-to-ground link channel change is deterministic, while the "noise" is random—atmospheric turbulence, weather changes, and ground scattering. These factors are superimposed on the deterministic skeleton to form the final channel state.

This discovery opened up a whole new world of ideas for Zuo Cheng.

The key to prediction is not to use a universal model to rigidly fit all changes, but to treat the "skeleton" and "noise" separately—the skeleton is calculated precisely using orbital mechanics, while the noise is estimated and compensated for in real time using its adaptive tracking algorithm. Each part performs its own function, and together they form a complete prediction scheme.

He named this idea the "two-layer prediction architecture"—the bottom layer is deterministic prediction, and the top layer is stochastic compensation.

That very evening, he drew a rough architectural sketch and sent it to Fang Ze and Chen Hao.

Ten minutes later, Fang Ze replied with a message: "The underlying orbit calculations require high-precision astronomical data and atmospheric refraction models. This part of the computation is quite extensive. Do you plan to run it on the ground terminal or on the base station side?"

"Good question." Zuo Cheng thought for a moment and replied, "On the base station side. Terminals have limited computing power, so track calculations are done at the base station, and the results are sent to the terminal via the control channel. The terminal only handles upper-layer adaptive compensation, resulting in less computational pressure and better real-time performance."

Fang Ze replied with a "reasonable".

Chen Hao's feedback came later, but it was more detailed—he pointed out that a synchronization mechanism is needed between the underlying orbit prediction and the upper-level adaptive compensation; otherwise, the time references of the two layers will not be aligned, and the prediction results will be biased.

Zuo Cheng noted this problem down in his notebook, marking it with a red asterisk. The synchronization mechanism is a critical detail; if not handled properly, it can become a bottleneck for the entire architecture.

I started writing the proposal on the fourth day.

By the afternoon of the second day, Zuo Cheng was stuck on a mathematical problem—the upper-layer adaptive compensation module required a set of initial parameters, which had to match the trajectory prediction results of the lower layer for the two layers to work together. However, the output of trajectory prediction was position and velocity, while the input of adaptive compensation was the channel state. The mapping relationship between the two involved a complex set of radio wave propagation equations, which Zuo Cheng struggled to work through for a long time.

As he rubbed his temples on the table, a voice came from beside him.

"Are you stuck on the propagation equation?"

Zuo Cheng looked up and saw Lin Ke. She was standing next to his workstation, holding a cup of coffee, her gaze sweeping over the derivation process spread out on his desk.

"Hmm. The mapping from orbital parameters to channel states is separated by an atmospheric propagation effect, and I'm not very familiar with ionospheric refraction." Zuo Cheng didn't hide anything.

Lin Ke put down her coffee, pulled up a chair and sat down next to him, then picked up a pen and wrote a series of formulas on the blank space of his draft paper.

"The ionosphere's main effects on signals are Faraday rotation and group delay. Faraday rotation can be approximated by this model—" she wrote quickly, her handwriting messy but logically clear, "the key parameter is the total electron density. Blue Bay Communications has real-time ionospheric monitoring data in its database; you can just call it up. Correcting group delay is even simpler; it can eliminate over 99% using a dual-frequency observation method."

Zuo Cheng looked at the formula she wrote and quickly compared it with his own derivation in his mind.

pass.

The effects of ionospheric refraction are concisely encapsulated in Lin Ke's model into two computable correction terms, which can be directly embedded into the propagation equation. He was previously stuck because he tried to derive the entire physical process of the ionosphere from scratch, which led him astray—in reality, such a precise physical model is not needed in engineering; a validated empirical model is sufficient.

"Thanks," Zuo Cheng said, looking at her. "You've been a huge help."

"It's mutual." Lin Ke stood up, took back her coffee, and said, "I glanced at the sketch of that two-layer prediction architecture on your desk. The idea of ​​using orbital mechanics for deterministic prediction in the lower layer is very clever. My channel modeling requires a time-varying channel prediction input. Once your solution is finalized, we might need to interface with it."

"No problem, we can connect anytime."

Lin Ke nodded and left.

Zuo Cheng watched her retreating figure, mentally raising his opinion of her. Lin Ke was straightforward, but her straightforwardness was reasonable—she wasn't helping out of politeness, but because she saw that their research directions overlapped, and establishing a technical cooperation relationship in advance would benefit both parties.

Such people are a scarce resource in companies.

After solving the problem of ionospheric refraction, the progress of the solution accelerated significantly. Zuo Cheng completed the full theoretical derivation of the two-layer prediction architecture in three days, and then spent another two days setting up the framework code in the simulation environment.

Chen Hao designed and implemented the synchronization mechanism through remote collaboration—he used a timestamp-based alignment scheme that was simple yet robust, and the time reference deviation between the two layers was controlled at the microsecond level.

On the eleventh day, Zuo Cheng produced the first set of simulation results.

The channel prediction accuracy of the two-layer prediction architecture under standard orbital conditions exceeds that of Blue Bay Communications' existing solutions by 42%.

Prediction lead time: It can predict channel state changes two seconds in advance with an error of less than 5%.

Two seconds. In satellite communication scenarios, a two-second advance prediction means that the system has enough time to adjust the transmission power, switch beam direction, and even prepare to switch to the next satellite in advance.

This data has exceeded the basic requirements of step four.

But Zuo Cheng did not stop.

Because standard orbital conditions represent only the simplest scenario. The real test lies in extreme conditions—switching strategies when faced with torrential rain, solar storms, ionospheric disturbances, and the simultaneous visibility of multiple satellites.

These scenarios need to be tackled one by one, as each one may harbor deadly pitfalls.

He closed his notebook and glanced at the calendar on the wall.

Day 11 of internship. Thirty-one days left.

There is enough time, but it cannot be wasted.

The light screen flickered silently in my consciousness:

[Main Quest Chain - Breaking the Communication Barrier - Stage Four Progress: 38%]

Zuo Cheng turned off the panel, picked up a pen, and wrote down a list of extreme scenarios he needed to tackle next in his notebook.

There are seven in total.

He circled the name of the first scene—"Rainfall Decay".

Start with the hardest one.