Scaling Multi-Core Fiber Without the DSP Headache
The capacity wall is real. Multi-core fiber is the way through it.
Long-haul networks are running out of room. Single-mode fiber is approaching its capacity ceiling, and operators are looking for the next step — one that doesn't mean trenching new cables.
Coupled multi-core fiber (MCF) is one of the most promising answers. By packing multiple cores into a single cladding, it multiplies capacity in the same physical footprint, with lower transmission loss than alternatives.
But there's been a catch.
The hidden cost: receiver complexity
Coupled MCF cores don't stay neatly separated. Signals leak between them — what engineers call crosstalk — and untangling that leakage at the receiver has historically required heavy, power-hungry digital signal processing.
The standard approach (joint MIMO equalization) has two problems operators feel directly:
• It's expensive. More processing means more silicon, more power, more cost per port.
• It's unreliable under strong coupling. The training algorithm can fail to converge, leading to dropped channels and operational headaches.
For many operators, this DSP burden has been the reason coupled MCF stayed on the roadmap rather than in the network.
A smarter way to equalize
Our team at MIMOPT developed a two-stage MIMO equalizer that rethinks the problem. Instead of solving everything in one massive computational block, we split the work into two focused stages — each doing what it does best.
The result, validated on a realistic 1000 km, 4-core, 16-QAM transmission link:
Metric | Improvement |
Receiver complexity | 66% lower than joint equalization |
Signal quality (SNR gain) | +2 dB under strong coupling |
Training reliability | Near-zero failure rate (vs. up to ~18% with conventional methods) |
In plain terms: a receiver that's cheaper to build, easier to operate, and more reliable in the field.
What this means for your network
Lower cost per bit.
A 66% reduction in DSP complexity translates directly into smaller, cooler, less expensive receiver hardware — at exactly the moment hyperscale and telco operators are pushing back hard on power and footprint budgets.
More headroom.
A 2 dB SNR improvement is real margin you can spend on reach, capacity, or resilience.
Production-grade reliability.
Eliminating the convergence failure mode means coupled MCF moves from "interesting research" to "deployable technology."
Future-proof scaling.
The architecture scales cleanly as core counts grow — so the savings get bigger, not smaller, as the technology matures.
From research to deployment
This work was presented at SPPCOM 2025, and it reflects how we operate at MIMOPT: we don't just publish results, we build the technology behind them. Our focus is helping operators and system vendors bring next-generation optical transmission out of the lab and into live networks.
Coupled MCF is no longer waiting on a breakthrough. It's waiting on partners ready to deploy it.
Key takeaways
• Coupled MCF is the most promising path past today's single-mode capacity limits.
• Receiver complexity has been the main barrier — until now.
• Our two-stage equalizer cuts complexity by 66%, adds 2 dB of margin, and removes the reliability problem.
• The technology is ready for system integration today.