210: Chapter 210 Three Hundred Guns Competing for a Target: Who Will Fire First?

sentry ship arrival countdown, 79 days remaining.

Pangu Laboratory, B2 Level, 10:20 AM.

When Academician Li Zhengyang returned after eating noodles, Su Che had already completed most of the underlying framework for the quantum fire control system.

He flipped through the technical documentation Su Che had pushed over from the beginning.

"I understand the core of the quantum precision positioning—extracting four-dimensional vectors from the radar point cloud: position, velocity, acceleration, and steering intent. The first three are easy to extract, but how do you extract the fourth?"

"Steering intent isn't measured; it's calculated."

Su Che brought up the raw detection data format of the Quantum Radar on the screen, "By feeding the rate of change of the target's acceleration vector over three consecutive frames into the main god ai's short-term prediction model, it outputs the most likely steering direction and magnitude within the next 0.5 seconds."

"Is a 0.5-second prediction window enough?"

"It's enough in deep space. There's no resistance in a vacuum. A target needs to fire propellants to change its orbit, and there's a physical delay from ignition to attitude change. 0.5 seconds can cover the vast majority of maneuvers."

Academician Li Zhengyang closed the document and set it on the left side of the test bench.

"I'll write the four-dimensional vector extraction; you just need to provide the interface for the short-term prediction model."

Su Che sent the interface definitions to the public encrypted warehouse.

"Academician Qian, you provide a version for the dimensionality reduction and compression of the point cloud data on the radar end."

Academician Qian Zhenhua replied with one word from his workstation on the east side: "Okay."

Three lines of work started simultaneously.

2:00 PM.

Zhao Mingyuan wandered over to the main console with a cup of plain water.

"New task?"

"Quantum fire control project."

He leaned over to the public big screen and watched for two minutes.

"Regarding cross-domain target locking, with three hundred Xuannv units, White Emperor units, and Mingdi Interceptors all squeezed into one fire control network to distribute targets, what allocation strategy are you planning to use?"

Su Che was writing the scheduling framework for multi-weapon cooperative fire control and didn't stop.

"Optimal shooting position priority. Each target is assigned only to the unit with the best shooting angle, shortest distance, and optimal weapon status."

"Sounds simple."

Zhao Mingyuan set his cup on the small table below the public big screen.

"But what if two Xuannv units have identical shooting conditions for the same target? Same distance, same angle, same weapon status—who gets it?"

He continued.

"With three hundred swarm units scattered across different orbits, the probability of identical shooting conditions is very low, but not zero. No matter how low the probability, when multiplied by a sufficient number of targets, it will definitely happen."

"You have an idea?" Su Che looked at him.

"Add a layer of random disturbance. When conditions are identical, the system assigns a tiny random weight to each of the two units. The higher one gets it. The random disturbance is small enough not to affect tactics, but large enough to break the deadlock."

Su Che tilted his head and thought about it.

"No need for randomness; use the ID number. When conditions are identical, the one with the smaller ID takes priority. Deterministic rules are more predictable on the battlefield than random rules; wingmen can anticipate their partner's behavior."

"Fine, you're the boss. But I need to add a rule on the Cognitive Graph side: if priority by ID leads to a specific Xuannv unit being assigned more than three targets consecutively, the fourth is automatically yielded to the next unit. This prevents the leading ID airframes from overloading."

"Add it."

Zhao Mingyuan stretched, picked up his cup, and returned to his workstation.

He didn't have a new task, but the swarm Cognitive Graph needed to be compatible with the fire control network, so this work was unavoidable.

sentry ship arrival countdown, 78 days remaining.

B2 Level, morning.

Academician Qian Zhenhua finished running the first version of the dimensionality reduction and compression algorithm for the radar point cloud data.

The raw point cloud data volume was huge; the data stream generated by the Global Deep Space Quantum Radar every second was at the terabyte level.

The fire control system needed to extract the target's four-dimensional vector from it in real-time; it couldn't handle it without compression.

Academician Qian Zhenhua's compression plan was layered filtering.

The first layer removed all echoes from background stars and known celestial bodies, cutting the data volume by ninety percent.

The second layer performed motion feature clustering on the remaining echoes, archiving stationary objects and keeping only moving targets.

The third layer performed high-precision tracking on moving targets, outputting a temporal feature matrix for each target.

After three layers, the data volume was compressed from terabytes to gigabytes, allowing the fire control system to process it in real-time.

However, Academician Li Zhengyang ran into issues when running the four-dimensional vector extraction using the compressed data in the weapon adaptation zone.

"Academician Qian, the distance threshold you used for the second-layer clustering is too large."

He posted two sets of data side-by-side on the public big screen.

On the left was the raw point cloud before compression; on the right was the compressed version.

"Here, two targets are four hundred kilometers apart. Under your clustering threshold, they were merged into one. The velocity vector extracted for the four-dimensional vector is the weighted average of the two targets, not the true value of either one."

Academician Qian Zhenhua walked over to look at that set of data.

"Four hundred kilometers is too close on a deep space scale. My threshold was set at the million-kilometer level."

"The spacing of formation targets is on this order of magnitude. If the opponent is also a swarm, and hundreds of them fly over in close formation, your clustering will smear them all into a blob."

Academician Qian Zhenhua didn't immediately refute him.

He returned to his workstation and changed the distance threshold of the clustering algorithm from a fixed value to an adaptive value. Areas with high target density automatically reduced the threshold, while areas with low density maintained a large threshold.

He modified it and reran it.

Two targets with a four-hundred-kilometer spacing were correctly separated.

But a new problem arose: when the number of targets surged to over a thousand, the computational cost of the adaptive threshold increased sharply, and the real-time performance dropped from millisecond-level to hundred-millisecond-level.

"A hundred milliseconds won't do." Su Che looked away from the scheduling framework code and glanced at the latency data. "The allocation cycle of the fire control network is ten milliseconds. You are ten times slower than it; the fire control system will frequently receive outdated data."

Academician Qian Zhenhua took a pen out of his pocket and drew a data flow topology diagram on a piece of scratch paper.

"I won't change the algorithm; I'll change the architecture. Clustering won't run on the radar end; it will be distributed to the onboard chips of each Xuannv unit for local computation. Each Xuannv unit will only be responsible for high-precision clustering of targets within its own field of view, and the results will be aggregated into the fire control network. With distributed computing, the latency will return to the microsecond level."

"Do the onboard chips of each Xuannv unit have any remaining computing power?" Academician Li Zhengyang asked from the other side of the weapon adaptation zone.

Zhao Mingyuan raised a hand at his workstation.

"Yes! The normal operation of the Cognitive Graph only consumes forty percent of the onboard computing power; the rest is always idle. Allocating fifteen percent for local clustering is more than enough."

Academician Qian Zhenhua wrote a version of the interface definition for the distributed clustering scheme and sent it to Zhao Mingyuan for onboard adaptation.

Zhao Mingyuan took it, and twenty minutes later replied with a message: "Connected. Ran a stress test with 1,200 targets; clustering latency is eight microseconds."

Eight microseconds.

More than a thousand times faster than the fire control network's ten-millisecond allocation cycle.

"Approved." Academician Qian Zhenhua locked the plan.

Afternoon.

Su Che wrote the multi-weapon cooperative fire control scheduling framework up to the core module.

Global scheduling for both particle beam and laser modes.

The problem wasn't in the technical details, but in the tactical logic.

Three hundred Xuannv units fighting simultaneously: some used particle beams for long-range targets, while others used lasers for short-range or atmospheric targets.

The Deep Space Main God AI had to decide in real-time which mode each unit should use.

After building the decision tree, Su Che ran a simulation.

The results didn't look right.

When the target distance was in the overlapping effective range of the particle beam and laser, the main god ai switched modes frequently.

One Xuannv unit switched three times in two seconds: particle beam - laser - particle beam. Although there was no dead zone with each switch, the charge/discharge cycle of the energy pre-storage capacitor was squeezed very tightly.

Academician Sun Qiwen walked over from the U-shaped array area and looked at the energy cycle log.

"Switching three times in two seconds, the capacitor lifespan will shrink from 100,000 times to under 20,000 times; it will be scrapped after one battle."

Su Che added a hysteresis interval to the switching conditions of the decision tree.

After the target entered the overlapping interval, the current mode was maintained without switching until the target completely left the overlapping interval and entered the exclusive interval of another mode, at which point switching was allowed.

Rerun.

The switching frequency dropped from three times every two seconds to at most once every fifteen seconds.

Academician Sun Qiwen looked at the cycle log again, said nothing, and walked away with his enamel mug.

sentry ship arrival countdown, 77 days remaining.

B2 Level, morning.

Multiple lines converged.

Academician Li Zhengyang's four-dimensional vector extraction module was locked, Academician Qian Zhenhua's distributed clustering scheme was locked, Su Che's multi-weapon cooperative scheduling framework was locked, and Zhao Mingyuan's Cognitive Graph compatibility adaptation was locked.

Su Che connected the four modules and fed them into the main god ai's comprehensive fire control simulator.

Three hundred virtual Xuannv units, plus twelve White Emperor units, plus six Mingdi Interceptors.

Multi-platform mixed formation.

Enemy simulation: 1,200 heterogeneous targets, including swarm formations, single large warships, and high-speed maneuvering small platforms, appearing in a mix.

Run.

The fire control network topology on the main screen was as dense as a spider web. Each target was uniquely assigned to one piece of equipment; no repetitions, no omissions.

The four-dimensional vector steering intent prediction had three deviations when the swarm formation changed, but the fire control network's ten-millisecond refresh cycle corrected it in the next allocation frame, without affecting the final hit.

Cross-domain target locking had a 0.8-millisecond allocation gap in the handover area between the White Emperor and the swarm, but Zhao Mingyuan's ID priority rule caught it.

The multi-weapon cooperative scheduling did not trigger any abnormal switching outside the hysteresis interval throughout the process.

Simulation finished.

Comprehensive hit rate popped up.

99.1 percent.

0.7 percentage points higher than the 98.4 percent when the final version of the directed energy weapon ran alone.

Academician Li Zhengyang opened his notebook and wrote a line next to that number: Fire control project, first round of joint debugging.

Su Che didn't lock the version immediately.

"Run another round, increase enemy targets to 3,000, and add an electromagnetic interference environment."

The simulator reloaded.

3,000 targets.

Global electromagnetic interference noise injected.

The Quantum Radar's signal-to-noise ratio dropped by twelve percent, and a large number of false echoes appeared in the point cloud data.

Academician Qian Zhenhua's distributed clustering experienced twenty-seven mis-clusterings under the interference of false echoes, mistaking false targets for real ones.

The fire control network assigned shooting resources to twenty-seven non-existent targets.

Hit rate: 97.3 percent.

Dropped by nearly two points.

"False echo filtering wasn't done." Academician Qian Zhenhua stood up from his workstation.

"It's not your job; it's a radar-side issue."

Su Che pulled out the feature data of the false echoes.

"But it's too late to change the radar now. Add a verification layer on the fire control side. After a target is locked, add a quantum entanglement state confirmation before firing. The quantum echo characteristics of real targets are fundamentally different from false echoes; entanglement state confirmation can filter out the fake ones in nanoseconds."

Su Che inserted an entanglement state verification node into the scheduling framework.

Rerun.

3,000 targets, global electromagnetic interference.

All twenty-seven false targets were filtered out before firing; not a single round of ammunition was wasted.

Hit rate returned to 99 percent.

Su Che archived the results of both rounds.

"Three days left, continue to polish the boundary conditions."

In the corner of the main screen, the sentry ship arrival countdown.

77 days.