In the spring of 2016, a Hsiung Feng III (雄風三型) anti-ship missile launched from a Republic of China Navy frigate and struck a fishing vessel off the coast of Penghu. A crew member was killed. Three others were injured. The navy called it an accidental firing during a routine drill. The incident became an embarrassment, a tragedy, and — in the strange calculus of weapons development — an inadvertent proof of concept.
The Hsiung Feng III is designed to hunt warships. It flies at supersonic speed, skimming the sea surface to evade radar. Its onboard seeker is engineered to acquire and lock onto large metal targets in cluttered, electromagnetically complex environments. The fishing vessel it struck that morning was a small fiberglass hull — minimal radar cross-section, surrounded by sea clutter, essentially the worst-case target geometry for the system's guidance electronics.
It hit anyway.
"It was a tragic accident," said Ray Kai (雷凱), a senior manager at Zyxel Technology. "But it proved the precision of the seeker beyond any doubt."
What the public never learned from that incident — what has remained undisclosed until now — is that the radar signal processor inside the Hsiung Feng III's seeker head was not made by CSIST, the government institute that designed and built the missile itself. It was made by a networking company in the Hsinchu Science Park, 450 meters from TSMC.
What Actually Happens in the Last Ten Kilometers of a Missile's Flight?
The answer has everything to do with a component most people have never heard of.
A modern anti-ship or air defense missile typically covers most of its range under data-link guidance — directed from the launch platform toward the general vicinity of its target. But in the terminal phase, roughly the last ten kilometers, the missile is on its own. Its onboard radar activates. The seeker begins transmitting and receiving signals, calculating from each returning pulse the precise distance and bearing of the target. It must do this within a single pulse interval — there is no time for multiple measurements and averaging. The calculation must be correct the first time, every time, as the missile closes at hundreds of meters per second.
The component that performs this calculation is the radar signal processor: a compact, ruggedized module embedded in the seeker head. It processes raw radar returns and translates them into the targeting commands that steer the missile. In the weapons community, it is sometimes called the brain and eyes of the missile — the intelligence that determines whether years of development and millions of dollars of hardware actually hit anything.
For the domestically produced missile series fielded by Taiwan's military, those processors come from Zyxel Group (合勤科技) and its defense subsidiary Zyflex Technology (智勤科技).
"CSIST makes the missile," said Li Bing-qin (李炳欽), Zyflex Technology's general manager. "The critical electronics inside — that's us."
Why Would Recalibrating a Missile Be Dangerous — and How Did Zyxel Solve It?
The relationship between Zyxel and Taiwan's missile program did not begin with the seeker processor. It began with a more mundane engineering problem that, left unsolved, would have created serious maintenance headaches for the navy.
Older variants of the Hsiung Feng anti-ship missile used conventional frequency generators — components that produce the precise radio frequencies a radar system needs to operate. The problem with those generators is that they drift. Over time, thermal cycling and mechanical stress cause the output frequency to shift away from its calibrated value. For a commercial radar system, this is an inconvenience.
For a missile warhead, it is a genuine operational problem: to recalibrate the generator, the warhead has to go back to the factory. Disassembling a live missile warhead carries obvious risks.
Zyxel's solution was a real-time frequency lock synthesizer — a module that continuously monitors its own output and corrects for drift without requiring any external intervention. Once installed, the missile never needs to return to depot for frequency recalibration. Both the Hsiung Feng II and the Hsiung Feng III have been retrofitted with the device.
It was a quiet engineering achievement that most of Taiwan's defense establishment never noticed. But it established Zyxel's credibility inside CSIST and opened the door to more consequential work.
Which of Taiwan's Missiles Carry Zyxel Electronics — and Why Does Each One Work Differently?
The signal processor business grew from there. Today, Zyxel's radar signal processing technology appears across Taiwan's entire domestically produced missile inventory — not just the anti-ship Hsiung Feng family, but the Sky Sword (天劍) air defense series as well.
The Sky Sword II is Taiwan's primary beyond-visual-range air-to-air missile, originally developed for the Indigenous Defense Fighter. Since 2018, according to Li, the military has progressively expanded its deployment to sea-based and land-based configurations — the Hai Chien Ling (海劍羚) naval variant and the Lu Chien II (陸劍二) ground-based system. In exercises, these missiles have been tested against small drone targets. Their hit rates, Li said, have been consistently high.
"The target is very small," Li noted. "But the accuracy is there. We like to think we helped with that."
The processor architecture for the Sky Sword differs from the Hsiung Feng design in ways that reflect their different operational environments. The Hsiung Feng's sea-mode processor must contend with sea clutter — the radar returns bouncing off wave surfaces — while maintaining sensitivity to the relatively flat radar signature of a ship's hull. It uses a magnetron amplifier whose output frequency varies over time, which creates its own signal processing complications: the processor must calculate in real time what frequency the radar is currently operating at, and continuously retune the receiver to match.
"The frequency changes as the missile flies," Ray Kai explained. "Our processor has to calculate what that frequency is in real time, and match the receiver to it. That's what gives the Hsiung Feng its sensitivity."
The sensitivity, as the Penghu incident demonstrated, is considerable. A small fiberglass fishing vessel, with a radar cross-section a fraction of the warships the missile was designed to engage, was acquired and tracked to impact. The seeker did not fail. The question of whether the launch should have happened at all is a separate matter entirely.
How Do You Win an Electronic War Before the First Shot Is Fired?
Missile guidance is the most visible part of what Zyxel does for Taiwan's military. It is not the most strategically significant.
That distinction belongs to electronic warfare — a domain that shapes the outcome of modern conflicts before a single conventional weapon is fired. The principle, as Li frames it, is simple in concept and extraordinarily difficult in execution: make your own signals unjammable, and jam the enemy's..
Electronic warfare divides into two domains. In the communications domain, the primary technique is frequency hopping: cycling through a pseudo-random sequence of frequencies fast enough that an adversary's jamming system cannot lock onto any single one long enough to suppress it. A communications system that cannot be jammed is a communications system that keeps working when everything else has gone silent. Zyxel has been developing frequency-hopping and spread-spectrum technology for over a decade — longer, Li noted, than most of the companies now rushing into the drone and counter-drone market.
"If a drone doesn't have anti-jamming capability, it's essentially useless in a contested environment," Chu Shun-yi (朱順一) said. "We've had that technology for years."
In the radar domain, the techniques are more varied and, in some respects, more psychologically unsettling. A radar system can be deceived rather than simply suppressed. Send it a carefully crafted false return — a signal that mimics the reflection of a target at a location where no target exists — and the system that depends on it begins tracking a ghost. The operators watching their screens see an aircraft or a missile where there is none. They may engage it. They almost certainly neglect the real target.
This was among the opening moves in the recent American military operation against Iran: electronic warfare systems used to feed false targets to radar networks, creating windows through which strike aircraft could operate undetected.
"It's a competition between spear and shield," Chu said. "You're always trying to do two things at once — protect your own signals and corrupt the enemy's picture of reality."
If You Can't Test Against China's Radar, How Do You Prepare for It?
The answer, for Zyxel, is simulation — and the machine they built to do it outperforms anything available on the open market.
Zyxel has built that machine. Its target signal simulator, developed in collaboration with CSIST and already delivered in quantities exceeding twenty units — five of them allocated to the army — can generate radar signatures across an instantaneous bandwidth of 1 GHz and simulate target velocities up to Mach 100. The Mach 100 ceiling is not arbitrary: it corresponds to the terminal velocity profile of an intercontinental ballistic missile. The simulator can replicate the radar signature of one of the most demanding targets in the threat environment Taiwan faces.
**Lin Chia-wen (林嘉雯)**, Zyxel's senior production manager, noted that when the company first entered this market, Western export controls restricted the highest-performance signal simulation equipment at instantaneous bandwidths of 250 megabits per second. Zyxel started at 500 Mbps. Its current high-end processor reaches 1 GHz — performance that Lin said is not available from foreign commercial suppliers at any price.
"You can't buy this abroad," she said flatly. "We built it because we had to."
The operational logic of the simulator closes a loop. Zyxel's engineers use it to generate the kinds of jamming and deception signals that a PLA radar or missile system might produce. They then test their own signal processors — the ones going into Taiwan's missiles — against those simulated threats. A seeker that can maintain lock through a Zyxel-simulated electronic attack is a seeker that stands a better chance of maintaining lock through the real thing.
"You give the simulator a signal," Chu explained. "It generates everything a hostile radar might throw back at you — false targets, jamming, deceptive returns. And then your missile has to figure out which return is real and go after it. That's the whole game."
Just How Much of Taiwan's Defense Has Zyxel Quietly Built?
Across the decades, the list of systems to which Zyxel has contributed reads like a catalog of Taiwan's defense modernization. The Hsiung Feng II and III real-time frequency lock synthesizers. The Sky Bow III (天弓三型) air defense simulation system. The Sky Sword II extended-range radar signal processor. Phase shifter modules. Passive army radar systems. Army communications equipment. High-speed modulators and demodulators. Portable signal direction-finding systems. Radar signal processing simulators. Sky Bow frequency-hopping systems. Naval vessel radar. The Yung Ying (勇鷹) advanced jet trainer's communications system. Electronic warfare ECM systems. The advanced digital control module for the Chien Hsiang (劍翔) anti-radiation loitering munition.
That last item is worth pausing on. The Chien Hsiang is Taiwan's answer to a category of weapon that has drawn intense attention since its effective deployment in conflicts across the Middle East and Ukraine: the loitering munition designed to hunt and destroy radar systems. Zyxel's digital control module, originally developed in 2010, now flies in those aircraft. It is another instance of technology developed for one purpose finding its most important application in another, years later, in a strategic context that no one fully anticipated when the work began.
Is What Zyxel Builds Actually Enough to Defend Taiwan?
It is the question that hangs over everything in this series — and one that Chu Shun-yi answers without illusions.
Taiwan's military doctrine has shifted in recent years toward asymmetric defense — emphasizing systems that are difficult to target, expensive to neutralize, and capable of imposing disproportionate costs on an attacking force. Missiles are central to that doctrine: large quantities of mobile, precision-guided weapons that can threaten any amphibious force attempting to cross the Taiwan Strait.
Those missiles are only as good as their guidance electronics. And those electronics, in large measure, come from a building in Hsinchu that most people walk past without a second glance.
"The defense situation here is difficult," Chu said, without particular drama. "Sometimes you can't import what you need. Sometimes what you're capable of building runs into obstacles that have nothing to do with technology. It's not a simple business."
He paused. "But we keep going."
（Related：
Profile | Taiwan's Chip, the KMT's Blind Spot, and the Cassandra Calling From Capitol Hill
｜
Latest
）
