6G Visions Demand Radical New Technologies

The 6G Promise That 5G Couldn't Keep

Cheng-Xiang Wang, Xiaohu You, Xiqi Gao, and Xiuming Zhu have spent years staring at the gap between what wireless engineers promise and what they deliver. Their 2023 survey in IEEE Communications Surveys & Tutorials—cited nearly 1,900 times—lays out the brutal truth: 5G is already hitting walls. Latency still lags behind what autonomous factories need. Bandwidth buckles under stadium crowds. And the network architecture itself, built for a world that streams video, cannot handle what comes next.

The authors argue that 6G must be something fundamentally different. Not a faster 5G. Not incremental improvements. A radical break. They identify seven key performance targets that no existing technology can meet. And they name the specific inventions—some barely past the lab stage—that might get us there.

What 5G Actually Failed to Do

The paper opens with uncomfortable numbers. 5G promised peak data rates of 20 Gbps. It delivers, in real conditions, far less. Latency targets of 1 millisecond remain elusive outside controlled test environments. Connection densities of 1 million devices per square kilometer—the number needed to make smart cities work—have not materialized in any commercial deployment (Wang et al., 2023).

The authors list 6G requirements that make 5G's promises look modest. Peak data rates of 1 terabit per second. Latency under 0.1 milliseconds. Connection densities of 10 million devices per square kilometer. Positioning accuracy within one centimeter. Energy efficiency 100 times better than 5G.

These numbers are not pulled from a wish list. They come from specific use cases. Telemedicine that lets a surgeon in Tokyo operate on a patient in Nairobi. Holographic meetings where remote participants feel physically present. Digital twins of entire cities that update in real time. Autonomous vehicle fleets that coordinate at highway speeds.

The Terahertz Problem No One Solved

Most of 6G's promised speed comes from moving to higher frequencies. Terahertz bands, above 100 gigahertz, offer enormous bandwidth. But Wang and his colleagues explain why this creates a physics nightmare.

Terahertz waves barely travel. They get absorbed by air, scattered by rain, blocked by walls. A 5G millimeter wave signal might reach 500 meters. A terahertz signal struggles to go 50 meters indoors. The authors document testbed results showing path loss that increases quadratically with frequency—meaning each jump in speed costs exponentially in range (Wang et al., 2023).

The proposed solution sounds like science fiction: intelligent reflecting surfaces. Thousands of tiny, programmable mirrors that bounce signals around obstacles. The paper describes prototypes using metamaterials—artificial structures that manipulate electromagnetic waves in ways natural materials cannot. These surfaces can steer beams in microseconds, creating virtual pathways where none exist.

But the authors are blunt about the gap between lab and deployment. "Current intelligent reflecting surfaces have limited bandwidth and efficiency," they write. "Their control algorithms require significant computational resources, and integration with existing network protocols remains an open challenge."

Why Your Phone Cannot Handle 6G

Here is the uncomfortable truth the paper reveals: 6G will demand hardware that does not exist yet.

Current smartphones use silicon-based chips. Silicon works fine for 5G. For terahertz frequencies, it becomes useless. The authors explain that terahertz signals require compound semiconductors—gallium nitride, indium phosphide, silicon germanium. These materials are expensive, difficult to manufacture, and generate enormous heat.

The paper reviews testbed results from multiple research groups. One prototype transceiver operating at 140 GHz achieved data rates of 100 Gbps. But it consumed 50 watts of power. A smartphone battery holds about 15 watt-hours. You would drain it in 18 minutes.

Wang and his colleagues identify three hardware breakthroughs required for 6G:

• ▸Terahertz transceivers with power efficiency below 1 picojoule per bit
• ▸Antenna arrays with thousands of elements that fit in a handset
• ▸Reconfigurable intelligent surfaces that switch states in nanoseconds

None of these exist commercially. The authors note that "significant research efforts are still needed to develop cost-effective and energy-efficient hardware solutions."

The Network That Learns Like a Brain

The most provocative section of the paper describes 6G's proposed architecture: a network that functions less like a traditional telecom system and more like a biological nervous system.

Current networks are hierarchical. Data flows from your phone to a tower to a central office to the internet and back. This creates latency. It wastes bandwidth. It makes the network brittle.

The authors propose a distributed architecture where intelligence lives at the edge. Base stations become mini data centers. They process data locally, make routing decisions, and coordinate with each other without a central controller. The paper calls this "semantic communication"—where the network understands the meaning of data, not just the bits.

A concrete example: In a 5G network, if you send a video, the network transmits every pixel. In a 6G semantic network, the system recognizes that you are sending a person talking. It transmits only the facial expressions, mouth movements, and voice data needed to reconstruct the video. The authors cite testbed results showing 90 percent bandwidth reduction for video calls using this approach (Wang et al., 2023).

But semantic communication requires artificial intelligence embedded at every network node. The paper acknowledges that training these AI models, ensuring they work across different languages and contexts, and protecting them from adversarial attacks remain open problems.

What the Paper Does Not Prove

Wang and his colleagues are careful to distinguish between what is possible and what is proven. Their survey covers over 300 papers and dozens of testbeds. But they flag several areas where the evidence is thin.

The terahertz channel models used in most simulations assume ideal conditions. Real environments have moving objects, weather changes, interference from other devices. The authors note that "existing channel models for terahertz bands have not been validated against extensive measurement data."

The energy efficiency gains claimed for intelligent reflecting surfaces come from simulations, not hardware. When you account for the power needed to control thousands of reflecting elements, the net savings shrink.

The semantic communication approaches work in controlled lab settings with limited vocabularies. Scaling them to the full complexity of human communication remains unproven.

And the 6G timeline itself is uncertain. The paper notes that ITU-R expects to reach a consensus on 6G vision by mid-2023. But standards development typically takes five to seven years. Commercial deployment would not begin until 2030 at the earliest. By then, the requirements may have shifted.

The Testbeds That Prove It Might Work

The paper's most valuable contribution may be its catalog of existing 6G testbeds. These are not simulations. They are physical hardware, operating at real frequencies, transmitting real data.

One testbed at Southeast University in China operates at 220-330 GHz. It achieved a data rate of 100 Gbps over a distance of 10 meters. The authors describe the setup: a transmitter with a 20 dBi horn antenna, a receiver with a 25 dBi antenna, and a channel sounder that measures propagation characteristics in real time.

Another testbed at Beihang University uses orbital angular momentum multiplexing—a technique that twists radio waves into helical shapes, each carrying independent data. The team demonstrated 32 Gbps over 100 meters.

A third testbed in Finland integrates artificial intelligence directly into the radio hardware. The system learns to optimize its transmission parameters based on real-time channel conditions. Early results show 30 percent improvement in spectral efficiency over conventional approaches.

The authors document these testbeds with the precision of engineers. They list frequencies, antenna configurations, modulation schemes, and error rates. This is not hype. It is evidence that specific technologies work, at least in controlled settings.

What This Actually Means

• ▸The terahertz band is real but limited. Expect 6G to use it only for short-range, fixed links—think wireless fiber replacement, not mobile phones. The first commercial 6G devices will likely use sub-terahertz frequencies around 100 GHz, not the 300 GHz bands that promise terabit speeds.

• ▸Intelligent surfaces will change indoor coverage. If the control algorithms mature, buildings could have walls that actively reflect signals to where they are needed. This would solve the coverage problem that plagues 5G millimeter wave. But the surfaces need to cost less than $10 per square meter to be viable.

• ▸Semantic communication will arrive first in specialized applications. Factory automation, where the vocabulary is limited and predictable, will benefit before general-purpose video calls. The bandwidth savings are real but the AI overhead is high.

• ▸Hardware is the bottleneck, not algorithms. The paper makes clear that we know how to process terahertz signals. We do not know how to do it efficiently. Expect significant investment in gallium nitride and indium phosphide manufacturing over the next five years.

• ▸The 6G timeline is 2030, not 2025. The testbeds prove feasibility but not manufacturability. The standards process has not even begun. Anyone promising 6G smartphones before 2028 is selling something that does not exist.