Rebirth: Starting from Lighting Up the Tech Tree

Chapter 29 Torrential Rain

Heavy rain attenuation is a well-known and long-standing problem in the field of satellite communications.

The reason is simple—raindrops absorb and scatter electromagnetic waves, and the higher the frequency, the greater the loss. The Celestial constellation uses the Ka band, which happens to be one of the most sensitive frequency bands to rainfall. A heavy downpour can cause the signal strength of the satellite-to-ground link to decrease by more than ten decibels, equivalent to instantly turning a highway into a muddy path.

Even more challenging is the extremely uneven spatial distribution of the rainstorms. Within the same area, it might be pouring rain in the east while the west enjoys clear skies, resulting in vastly different channel conditions for two ground terminals less than five kilometers apart. This spatial inconsistency renders uniform compensation strategies virtually ineffective.

Zuo Cheng searched through Lanwan Communications' actual measurement database and found that the data under heavy rain conditions was pitifully scarce—only eleven sets, and all of them were collected from the same rainfall event, with a very narrow spatial coverage.

Insufficient data. This is the same problem again.

But this time he couldn't carry his equipment to collect data on-site like last time—he couldn't just wait for a torrential downpour to go and wait. He needed a different approach.

On the twelfth night, Zuo Cheng lay facing the ceiling in his dormitory for half an hour, then suddenly sat up.

Meteorological data.

Heavy rain is a meteorological phenomenon, and Earth's meteorological monitoring network has accumulated decades of historical data. If we can combine the rainfall intensity distribution data from meteorological departments with the radio wave propagation model of satellite links, we can use historical meteorological data to "synthesize" a large number of heavy rain attenuation channel samples.

This is consistent with Chen Hao's previous approach to augmenting underground space data, but at a higher level—instead of simulating the signal environment of a closed space, it uses real meteorological observation data to drive a physical propagation model, and the resulting synthetic data is statistically closer to the real situation than pure simulation data.

That very evening, Zuo Cheng sent an email to Han Zhe requesting access to the meteorological data interface provided by Blue Bay Communications. Han Zhe approved it the following morning—the allocation of technical resources for the Sky Dome project was of very high priority.

After obtaining the meteorological data, Zuo Cheng spent two days building a "meteorological-driven channel synthesizer"—inputting the rainfall intensity distribution at any given time and geographical location, and outputting the satellite-to-ground link channel status under the corresponding conditions. The core of the synthesizer is a set of empirical formulas for Ka-band rain attenuation, and after calibration with measured data, the accuracy is controlled within an acceptable range.

With the synthesizer, he generated 20,000 sets of channel data under heavy rain conditions in one go, covering various combinations from light rain to torrential rain, from tropical to temperate zones, and from plains to mountains.

Then feed this data to the two-layer prediction architecture to run.

The result wasn't very good.

The 42% accuracy advantage under standard conditions is halved in heavy rain scenarios, dropping to only 19%. The reason is clear: the signal attenuation caused by heavy rain is too drastic, and the upper-layer adaptive compensation module cannot keep up with the speed of change, always lagging behind.

"The problem isn't the compensation speed." Zuo Cheng frowned at the simulation data for an entire afternoon before finally finding the crux of the matter: "The problem lies in the fact that there are absolutely no meteorological factors in the underlying deterministic predictions."

The bottom layer of the two-layer architecture is pure orbital mechanics—it can accurately predict the satellite's position and velocity, but it is unaware that it is raining heavily on the ground. The channel conditions predicted by the bottom layer are the "clear-day version," while the top layer receives an initial value with a large deviation, making compensation naturally difficult.

The solution is straightforward: add meteorological information to the underlying layer.

By integrating real-time rainfall data from ground weather stations into the underlying prediction module, the current weather conditions are considered during orbit prediction. The output is no longer a "sunny day version" but a "current weather version" of the channel prediction. The upper layer only needs to compensate for the remaining random errors, significantly reducing the workload.

But "direct" does not mean "simple".

The update frequency of meteorological data differs from the processing frequency of satellite signals by several orders of magnitude—weather stations update data every five minutes, while channel estimation requires milliseconds of response time. Five minutes is a century for a communication system. How can the gaps between two weather updates be filled?

Zuocheng was stuck for another day.

On the afternoon of the sixteenth day, he ran into Tang Xu on his way to the cafeteria to get food.

Tang Xu was carrying a plate of braised eggplant and a bowl of rice. He saw Zuo Cheng take the initiative to greet him, hesitated for a moment, and then sat down opposite him.

"Zuo Cheng, I have something I want to ask you." Tang Xu scratched his head, a hint of embarrassment hidden beneath his憨厚 (honest and simple) expression. "The beamforming algorithm for my antenna array requires channel state information as input, but the existing channel estimation scheme provides data refresh rates that are too low. I'd like to increase the refresh rate, but I'm afraid it will increase the computational burden. Is it possible for your prediction architecture to provide a high refresh rate channel prediction interface for me to use?"

Zuo Cheng stopped picking up food with his chopsticks.

High refresh rate channel prediction interface.

Tang Xu's needs and the problem he is solving are two sides of the same coin—Tang Xu needs high-frequency channel prediction output, while Zuo Cheng needs to perform high-frequency interpolation compensation between low-frequency meteorological data.

If we consider meteorological data as a "low refresh rate external input", then the interpolation compensation mechanism he needs and the high refresh rate interface Tang Xu needs are essentially the same technical component.

"What refresh rate do you need for your beamforming?" Zuo Cheng asked.

"Ideally, in milliseconds."

"That's perfect." Zuo Cheng put down his chopsticks. "I have a problem now—the weather data update cycle is five minutes, but my forecasting module needs millisecond-level weather correction. I'm thinking about how to interpolate between two updates. If this interpolation mechanism is developed, it can simultaneously output the millisecond-level channel forecasts you need."

Tang Xu's eyes lit up immediately.

"Shall we do it together?" He put down his spoon, gesturing with his hands. "The core of interpolation is estimating trends between known data points, right? The beam scanning process of the antenna array itself generates a large amount of spatial sampling data, which can be used as an auxiliary constraint for meteorological interpolation—because the difference in signal attenuation received by beams in different directions inherently contains information about the spatial distribution of rainfall."

Zuo Cheng paused for a second.

He hadn't even considered this idea.

Using beam scanning data from the antenna array to assist in meteorological interpolation is equivalent to using the antenna as a simple "weather radar." There's no need to wait for the weather station to update every five minutes; the antenna's own observation data is refreshed every millisecond, which can completely fill in the gaps.

"Tang Xu, your idea is brilliant," Zuo Cheng said sincerely.

Tang Xu chuckled, revealing a row of neat white teeth: "Of course, we help each other out."

The two began working together that very afternoon.

Zuo Cheng was responsible for designing the mathematical framework of the interpolation algorithm, while Tang Xu was responsible for the extraction and preprocessing of antenna array beam data. Their technical directions were originally independent, but at this intersection, they precisely aligned.

Three days later, the joint solution was successfully implemented.

The meteorological interpolation module utilizes antenna beam scanning data to achieve millisecond-level updates of rainfall distribution, improving accuracy by four times compared to pure weather station data. Zuo Cheng embedded this module into the bottom layer of a two-layer prediction architecture and re-run simulation tests of heavy rain scenarios.

When the results came out, Tang Xu was standing behind Zuo Cheng watching the screen.

Channel prediction accuracy in heavy rain scenarios: exceeding 36% of Blue Bay Communications' existing solutions.

From 19% to 36%, it almost doubled.

Tang Xu whistled.

Zuo Cheng leaned back in his chair, a slight smile curving his lips.

The downpour is over.

There are six extreme scenarios remaining.

He opened his notebook, ticked "Rainfall attenuation," and then looked at the next name on the list—"Solar storm."

Out of the corner of his eye, he noticed Cheng Yuan raise his head from his workstation, his gaze behind his gold-rimmed glasses shifting from the data on the screen to the backs of Zuo Cheng and Tang Xu standing side by side.

What was in that gaze was not easy to discern.

It didn't feel like hostility; it felt more like a sense of urgency from being left behind.