Wavelength Division Multiplexing (WDM)

Wavelength Division Multiplexing (WDM) is the “prism” of fiber optics. It allows you to send multiple data streams over a single strand of fiber by giving each one its own “color” (wavelength).

The variations (CWDM, DWDM, LWDM, and the likely intended FWDM) exist because of a fundamental engineering tradeoff: How much data can you pack in before the equipment becomes too expensive or hot?

Quick Comparison Table

1. DWDM (Dense WDM)

DWDM is the “heavy lifter.” It packs channels so tightly (less than 1nm apart) that it requires high-precision, temperature-stabilized (cooled) lasers.

• Why use it? When you need massive capacity (Terabits) over long distances. It works perfectly with optical amplifiers (EDFAs), meaning the signal can travel hundreds of miles without being converted back to electricity.

2. CWDM (Coarse WDM)

CWDM is the budget-friendly version. Because the “lanes” are 20nm wide, the lasers don’t have to be precise. If the laser heats up and the wavelength drifts slightly, it’s still within its wide lane.

• Why use it? It’s the go-to for “Metro” networks (within a city) where you only need 8–16 channels and want to avoid the cost of cooling systems.

3. LWDM (LAN-WDM)

LWDM is a newer middle ground specifically designed for high-speed Ethernet (400G and 800G). It operates in the O-band (1310nm) because that area has “zero dispersion”—meaning the light pulses don’t spread out as much, which is critical for the ultra-fast speeds used in 2026 data centers.

• Why use it? It’s the standard for 400G-LR8 or 800G-LR transceivers used to connect buildings within a campus.

4. FWDM (Filter WDM)

FWDM isn’t a “grid” like the others. It uses thin-film filters to combine or separate three specific, widely spaced wavelengths—usually for Fiber-to-the-Home (FTTH).

• Why use it? It allows a single fiber to carry Internet (1310nm), Voice (1490nm), and CATV/Video (1550nm) to your house simultaneously.

Why so many variations?

It comes down to Physics vs. Economics:

1. Heat Management: Tight spacing (DWDM) requires power-hungry cooling. Wide spacing (CWDM) is “set it and forget it.”
2. Fiber Quality: Certain wavelengths (like the O-band used by LWDM) are better for high-speed, short-distance runs, while others (C-band used by DWDM) travel further with less signal loss.
3. Specific Use-Cases: 5G towers (MWDM), home internet (FWDM), and AI-driven data centers (LWDM) all have different requirements for power, cost, and speed.

Which is most common in 2026?

• Inside the Data Center (Intra-DC):LWDM is currently the dominant choice for high-speed 400G and 800G links between switches. For very short reaches within a rack, SWDM (Shortwave WDM) is also common on multi-mode fiber.
• Building-to-Building (Campus):CWDM is most common because it’s inexpensive and supports 10G/25G/100G easily over the few kilometers between buildings without needing expensive active cooling.
• Data Center Interconnect (DCI – Across a City):DWDM is the standard here. As AI training clusters now span multiple physical sites, the massive bandwidth of DWDM is required to link them together as if they were one giant computer.

While all these technologies can technically run over “single-mode fiber,” they have very different requirements regarding the grade and physical characteristics of that fiber.

Fiber Compatibility Table

Key Fiber Distinctions

1. The “Water Peak” Problem (Critical for CWDM)

Older single-mode fiber (G.652.A/B) has high impurities that absorb light at 1383nm. This is known as the “water peak.”

• The Conflict: CWDM uses the entire spectrum from 1270nm to 1610nm. If you use old fiber, your middle channels (1370nm–1410nm) will simply disappear.
• The Solution: Modern G.652.D fiber is “Low Water Peak,” allowing all 18 CWDM channels to work.

2. Dispersion Management (Critical for DWDM)

In DWDM, channels are so close together that they can physically interfere with each other through a phenomenon called “Four-Wave Mixing.”

• Standard Fiber (G.652): Is fine for many applications but may require dispersion compensation units (DCUs) at very long distances.
• NZDSF (G.655): This fiber was specifically engineered for DWDM. It introduces just enough dispersion to keep the channels from overlapping/interfering while still being efficient.

3. Bend Sensitivity (Critical for FWDM)

Since FWDM is the backbone of Fiber-to-the-Home (FTTH), it almost always runs on G.657 fiber.

• Unlike standard data center fiber, G.657 can be tied in a (loose) knot or stapled to a baseboard without the light leaking out. This is vital because home installations involve many tight turns that would “break” a DWDM signal.

4. The Multi-mode Exception (SWDM)

SWDM (Shortwave WDM) is the variant you will see most in Building/DC connections using Multi-mode fiber. It uses OM5 (Lime Green) fiber to send 4 wavelengths over one strand, allowing 40G or 100G speeds over fiber that usually only supports 10G.

Summary for Data Centers

If you are currently wiring a building or data center:

• For 400G/800G links: Use G.652.D or G.657.A2 (Single-mode) to support LWDM.
• For short rack-to-rack (cheap): Use OM5 (Multi-mode) to support SWDM.
• For Long-Haul/DCI: Stick to G.652.D and high-grade DWDM equipment.

To calculate whether your fiber link will actually work, you need to find the Safety Margin. If the number is positive (ideally above 3 dB), your link is healthy. If it’s negative, the signal won’t reach the other end.

Here is the step-by-step math used in the industry today.

1. Calculate the “Power Budget”

This is the total amount of signal “strength” you have to play with, determined by your hardware (transceivers).

Power Budget (dB = TX Power (min) – RX Sensitivity (min)

Example: You have a 100G CWDM4 transceiver.

• TX Power (min): -6.5 dBm
• RX Sensitivity (min): -11.5 dBm
• Power Budget: -6.5 – (-11.5) = 5dB

2. Calculate the “Total Link Loss”

This is the sum of every “speed bump” the light hits between point A and point B.

3. The Final Calculation (Safety Margin)

A good design includes a Safety Margin to account for fiber aging, dust on connectors, or future emergency repairs (extra splices).

Safety Margin = Power Budget – Total Link Loss

Practical Scenario: 10km CWDM Campus Link

• Power Budget: 10 dB (Standard 10G CWDM SFP+)
• Fiber Loss (10km): 10 x 0.35 = 3.5dB
• Mux/Demux Loss: 3.0 dB (one at each end = 6.0 dB)
• Connectors (4 pairs): 4 x 0.5 = 2.0dB
• Total Loss: 3.5 + 6.0 + 2.0 = 11.5dB

Result: 10dB – 11.5dB = -1.5dB (FAILED)

In this case, the link will likely drop packets or not link up at all because the Mux loss is too high for the distance. You would need a “High Power” (ER or ZR) transceiver.

Industry Tips for 2026

• The “Dirty Connector” Penalty: In 400G and 800G networks, a single fingerprint on an LC connector can add 2.0 dB of loss instantly. Always clean before you plug.
• Mux Stacking: If you stack a CWDM mux and a DWDM mux together (hybrid), you must add the insertion loss of both filters to your calculation.
• O-Band vs C-Band: Remember that 1310nm (LWDM/CWDM) loses strength faster than 1550nm (DWDM). You can’t use the same distance assumptions for both.

I have consolidated the technical specifications, fiber requirements, and economic metrics into three master comparison tables.

1. WDM Technology Master Comparison

This table compares the multiplexing technologies based on their spectral and hardware requirements.

*Requires optical amplification (EDFAs).

2. Fiber Compatibility & Performance

Choosing the right fiber is critical to ensure the “colors” of light aren’t absorbed or distorted over your 5km distance.

3. The 100G vs. 400G Economic Tradeoff

For your specific 5km link, here is the side-by-side comparison of running four legacy links versus one modern 400G link.

Don't be shellfish…

Twitter

Linkedin

Google

Facebook

Email

Diggit