Introduction

Optical fiber communication has revolutionized global telecommunications by offering massive bandwidth and low attenuation over long distances. However, a single optical carrier can only transmit so much data before fundamental limits (like fiber nonlinearity and amplifier bandwidth) are reached. Wavelength Division Multiplexing (WDM) emerged as a solution: by sending many signals at different wavelengths (colors of light) through the same fiber, network engineers can multiply the capacity of existing fiber infrastructure without laying new cables. In essence, WDM treats different optical wavelengths as independent communication channels that co-propagate in one fiber. As illustrated in Figure 1, multiple optical sources operating at distinct wavelengths are first combined by an optical multiplexer (MUX), transmitted simultaneously through a single optical fiber, and then separated again by a demultiplexer (DEMUX) at the receiver end. This fundamental configuration enables parallel data transmission over the same medium while maintaining channel independence and minimal crosstalk.

Early WDM systems began with just 2 wavelengths (e.g. 1310 nm and 1550 nm), but modern Dense WDM (DWDM) can pack dozens of channels in the same fiber. For example, combining 40 channels can multiply a fiber’s throughput fortyfold. Indeed, current DWDM systems support on the order of 80–100 wavelengths in the C-band, each at 10, 100, or more Gb/s, yielding multi-terabit per second totals. WDM technology is largely protocol-agnostic and can carry various signal types (SDH/SONET, Ethernet, etc.) in parallel. A 16-channel DWDM system essentially acts like 16 virtual fibers, each capable of transporting a signal such as STM-16/OC-48 (~2.5 Gb/s) — a legacy example used here for illustration, not as a definitive or exclusive reference. This dramatically increases capacity; for instance, multiplexing 32 OC-192 signals (10 Gb/s each) via DWDM gives an aggregate 320 Gb/s, and research demonstrations have reached beyond 1 Tb/s on a single fiber in the late 1990s. In practical networks, WDM (especially DWDM in the 1550 nm band) has been fundamental in scaling long-haul and metro optical links to meet the internet’s explosive growth in data traffic.

While WDM focuses on multiplexing in the wavelength domain, researchers have also turned to space-division multiplexing (SDM) to further break through fiber capacity limits. The capacity of a standard single-mode fiber is approaching about 100 Tb/s (100 Tbit/s) due to nonlinear Shannon limit constraints [2], leading to a potential “capacity crunch.” As illustrated in Figure 2, the nonlinear Shannon limit arises from the trade-off between optical signal power, amplified spontaneous emission (ASE) noise, and nonlinear distortions, which together constrain achievable information rates in fiber systems.

Before reaching this limit, various multiplexing strategies—such as WDM, coherent modulation, and advanced coding—were developed to push spectral efficiency and total throughput. However, as shown in Figure 3, even these techniques eventually approach the saturation region of standard single-mode fiber capacity, motivating the exploration of SDM as a breakthrough solution [3].

SDM offers new parallel channels by using multiple spatial paths in one fiber – either multiple cores or multiple modes. In recent years, SDM experiments have achieved petabit-per-second class data rates. For example, in 2025 a 19-core fiber transmitted 1.02 Pb/s over 1,808 km [5]. This article presents a concise technical overview of WDM (coarse and dense) and SDM technologies, outlining their operating principles, key components, and fiber bands. It also highlights current challenges—such as cost, complexity, crosstalk, and nonlinearity—and the engineering solutions addressing them.

Fundamentals of WDM

Principle of operation

Wavelength Division Multiplexing (WDM) enables multiple data streams to travel through a single optical fiber by assigning each to a distinct wavelength channel. Similar to a prism, a multiplexer combines various light wavelengths at the transmitter, while a demultiplexer separates them at the receiver. Each wavelength propagates independently with minimal interference when the fiber is linear and dispersion-managed, greatly increasing total capacity without adding fibers. Standardized wavelength grids ensure channel compatibility across systems. DWDM narrow spacing (50–100 GHz, ≈0.4–0.8 nm) within the C-band (1530–1565 nm) to support over 80 channels. CWDM employs wider spacing (~20 nm) across 1270–1610 nm, offering up to 18 channels with simpler, lower-cost components.

CWDM vs. DWDM: CWDM and DWDM cater to different network needs. CWDM supports up to 16–18 channels spaced about 20 nm apart across the O- to L-bands (1270–1610 nm). Its wide spacing allows the use of uncooled lasers and broadband filters, making it low-cost and tolerant to wavelength drift. CWDM is ideal for short- to medium-distance links, such as access, campus, or metro networks, where simplicity and cost-efficiency matter most. DWDM, in contrast, operates mainly in the C- and L-bands (1530–1625 nm) with tight channel spacing (50–100 GHz, ≈ 0.4–0.8 nm), enabling 40–96 wavelengths per fiber, each carrying 10–400 Gb/s—reaching up to 8 Tb/s. It relies on temperature-stabilized lasers and high-precision filters but delivers maximum capacity and reach, making it the backbone technology for long-haul and undersea communication systems where fiber resources are scarce but bandwidth demand is high.

WDM multiplexers and filters: Achieving WDM in practice relies on optical filters that can combine and separate specific wavelengths. Several technologies are widely used:

• Thin-film filters (TFF): These are multilayer dielectric interference filters deposited on glass. Each filter transmits one target wavelength band while reflecting others. By cascading multiple TFFs, a module can successively drop or combine channels (λ_₁, λ_₂, etc.), making this approach ideal for low-channel-count CWDM or DWDM units (typically 4–16 channels). TFF-based multiplexers are compact and cost-effective, though cascading many filters introduces accumulated insertion loss, limiting scalability beyond ~16 channels. Despite this, they remain popular for passive modules and add/drop filters due to simplicity and robustness.
• Arrayed waveguide gratings (AWG): An AWG is a planar lightwave circuit that functions like an integrated diffraction grating. It uses an array of waveguides with incrementally increasing path lengths to create wavelength-dependent interference at output ports. This allows simultaneous multiplexing or demultiplexing of dozens of channels (40–100 or more) with uniform loss. AWG devices form the backbone of DWDM terminals in metro, core, and submarine systems. They are fixed to a specific ITU grid, making them ideal for static MUX/DEMUX applications but less adaptable for reconfigurable optical add/drop multiplexers (ROADMs).
• Fiber Bragg gratings (FBG): An FBG consists of periodic refractive index variations written inside an optical fiber, reflecting a narrow wavelength band while transmitting others. Combined with an optical circulator, an FBG can selectively add or drop individual channels. These devices offer excellent spectral selectivity and minimal loss, making them useful for notch filtering, channel equalization, or limited-channel add/drop functions. However, implementing many FBGs for large channel counts is complex, so they are often combined with AWGs or tunable filters.

Overall, the choice of WDM filtering technology depends on system design — channel count, spacing, and flexibility. TFFs suit compact CWDM modules, AWGs power high-capacity DWDM systems, and FBGs provide precision filtering. All must ensure high channel isolation (30–50 dB) and low insertion loss to minimize crosstalk and maintain signal integrity.

Fiber transmission windows and wavelength bands

Modern single-mode silica fibers exhibit low-loss “transmission windows” spanning the telecom O, E, S, C, and L bands. In practice, system design maps services onto these bands according to loss, dispersion, and amplifier availability (e.g., EDFA in C/L, Raman options toward S). For a deeper primer on each band’s history, loss figures, and amplifier coverage, see our separate online article on fiber transmission bands. Figure 4 illustrates the fiber attenuation curve with shaded O–L regions and the legacy 1383 nm water-peak that historically limited the E-band.

When planning WDM channels, engineers choose bands and spacings to balance capacity and performance. CWDM typically places 8 or 16 channels from ~1270 nm to 1610 nm, broadly covering O through L. Early CWDM deployments often skipped the ~1383 nm water-peak unless low-OH (G.652.D) fiber was used. DWDM concentrates channels in C (and L) because mature amplifier ecosystems exist there. The ITU-T G.694.1 standard defines DWDM frequency grids (e.g. anchored at 193.10 THz with 50 or 100 GHz increments). In fixed-grid systems, each channel has a designated central frequency/wavelength and tight tolerance (laser drift must be controlled to within ~0.01 nm). Filters and DEMUX must have sharp roll-offs to prevent overlap. More recently, flexible grids and “gridless” networks have emerged, where channels can occupy variable bandwidth (especially for 400G+ superchannels), but that is beyond our current scope.

As data traffic keeps increasing, researchers are investigating multi-band WDM – essentially using all available fiber bands (O through L, possibly plus 850 nm window) to maximize fiber capacity. This requires multiple amplifier types (e.g. Raman amplifiers for S-band, TDFA for S, EDFA for C/L, etc.) and is an active R&D area. A fully multi-band system could theoretically utilize ~10 THz or more of spectrum (C-band offers ~4.5 THz, C+L ~9 THz, adding S could make ~13–14 THz). In laboratory experiments, transmission across O+E+S+C+L has been shown in pieces, but practical deployment will depend on economical amplifier solutions across those bands. In any case, exploiting more wavelengths and bands is one pillar of scaling optical network capacity.

Pushing beyond WDM: Space-division multiplexing (SDM)

After reaching the practical limits of wavelength-division multiplexing, the next major step in increasing fiber capacity is to transmit multiple parallel data streams through separate spatial channels within the same fiber — a concept known as Space-Division Multiplexing (SDM). Conventional single-mode fiber (SMF) supports only one spatial channel, whereas SDM introduces N independent channels that can operate in parallel. These channels can be realized primarily through multi-core fibers (MCF) and few-mode fibers (FMF), often in combination with wavelength-division multiplexing (WDM) and polarization multiplexing, resulting in a multiplicative capacity gain. SDM has emerged as a promising approach to surpass the approximate 100 Tb/s capacity limit of single-core fibers, supported by numerous high-capacity experimental demonstrations.

Multi-core fiber (MCF)

An MCF integrates multiple light-guiding cores within a single cladding of 125 µm diameter, ensuring compatibility with existing fiber infrastructure. Typical designs include 7-core and 19-core configurations arranged in a hexagonal pattern. The key technical challenge is inter-core crosstalk, which is mitigated using trench-assisted structures (low-index regions surrounding each core) or heterogeneous cores with slightly varied refractive indices to disrupt phase matching. These designs have achieved isolation levels as low as –30 to –70 dB/km, enabling long-distance transmission such as 1.02 Pb/s over 1,808 km using C+L bands with advanced MIMO signal processing. As illustrated in Figure 5, MCF design parameters — including core count, layout, refractive-index profile, outer cladding thickness (OCT), and cladding/coating diameters — directly influence optical performance and mechanical stability, requiring careful optimization for targeted applications.

In coupled-core MCFs, cores are intentionally close to allow strong coupling, forming super-modes across multiple cores. This enables denser packing but demands complex MIMO digital signal processing (DSP) to separate mixed signals, similar to wireless MIMO systems. A coupled 4-core fiber recently achieved 1.02 Pb/s over 1,000 km, demonstrating the power of hybrid optical–DSP optimization.

Few-mode fiber (FMF)

FMFs use a larger single core supporting several spatial modes (e.g., LP₀₁, LP₁₁, LP₂₁), each carrying independent data. Mode multiplexers and demultiplexers are employed for coupling and separation, while mode coupling and modal dispersion necessitate compensation via MIMO equalization. FMF demonstrations have reached over 2 Pb/s using 6 modes × 19 cores, confirming the scalability of SDM.

Benefits and trade-offs: SDM provides exponential capacity scaling by exploiting spatial diversity rather than additional wavelengths. MCFs are more compatible with existing systems, requiring mainly fan-in/fan-out devices, while FMFs demand complex receiver designs but promise high density for shorter links. Combining SDM with WDM — for example, 80 DWDM channels per core across multiple cores — has enabled record-breaking petabit-per-second transmissions, positioning SDM as a cornerstone of next-generation ultra-capacity optical networks.

Challenges in deploying advanced multiplexing systems

The implementation of Wavelength-Division Multiplexing (WDM) and Space-Division Multiplexing (SDM) in real optical networks introduces significant technical and economic challenges. These difficulties span cost, signal integrity, nonlinear effects, and system engineering.

Cost and system complexity

Expanding the number of transmission channels—whether by wavelength or spatial dimension—inevitably increases system complexity and capital cost. In WDM systems, dense channel spacing requires high-precision lasers, filters, and amplifiers, as well as dispersion and gain control. Although the component cost is higher, the cost per bit decreases because fiber bandwidth is more efficiently used.

For SDM, the economic challenge is steeper. MCFs remain expensive to fabricate in volume, requiring specialized preform and drawing processes. New components—such as multi-core connectors, fan-in/fan-out couplers, and amplifiers—must also be developed. A single MCF link may require several transmitters and receivers operating in parallel, unless advanced photonic integration consolidates them. FMFs face similar cost barriers, as they depend on complex optics and high-power digital signal processing (DSP) for mode separation and compensation. As integration improves, costs will fall, but early deployment will likely be limited to submarine systems or core backbones, where extreme capacity justifies expense.

Crosstalk and channel isolation

In WDM systems, wavelength-selective components typically achieve >30 dB isolation, minimizing linear crosstalk. The main interference source is nonlinear interaction between channels (discussed below). In contrast, crosstalk is intrinsic to SDM. For uncoupled MCFs, fiber design focuses on suppressing inter-core coupling using trench-assisted or heterogeneous-core structures, targeting average coupling below –30 dB. Coupled-core MCFs and FMFs, however, deliberately allow strong mode coupling, requiring MIMO-based DSP at the receiver to separate mixed signals. This approach succeeds only if the channels remain phase-coherent and the coupling matrix is well-behaved.

Another critical challenge lies in component-level isolation. Multi-core fan-in/fan-out devices must align each core precisely, and multi-core EDFAs must amplify all cores uniformly without optical coupling. Early multi-core amplifiers exhibited unequal gain and parasitic coupling, but newer prototypes have demonstrated <1 dB gain variation across 19 cores, showing that practical performance parity is achievable.

Fiber nonlinearities

As more channels and higher power levels are launched simultaneously, the Kerr nonlinearity in optical fiber induces distortions such as Four-Wave Mixing (FWM), Cross-Phase Modulation (XPM), and Stimulated Raman Scattering (SRS). FWM is particularly critical in DWDM systems, where interactions among closely spaced wavelengths generate new spectral components that can overlap existing channels. XPM introduces phase noise that converts to amplitude distortion through chromatic dispersion. These effects establish the so-called nonlinear Shannon limit, where increasing power or channels no longer improves capacity due to inter-channel interference.

Network designers mitigate nonlinearities by optimizing dispersion maps, adjusting channel spacing, and reducing launch power per channel as channel density rises. Advanced coherent receivers use DSP algorithms—such as digital backpropagation or machine-learning-based equalization—to compensate some of these impairments. However, these techniques are computationally intensive and energy-hungry. Research continues into new fiber designs, joint multi-channel processing, and optical phase conjugation to manage nonlinearities, but this remains a fundamental physical limitation of fiber communication.

Engineering and operational constraints

Engineering a high-capacity optical system requires maintaining precise wavelength control, power balance, and component alignment. In DWDM networks, laser wavelengths must remain locked to the ITU grid with tolerances near 0.01 nm. Temperature drift and optical power imbalance are managed using wavelength lockers and gain-flattening filters within amplifiers. Optical monitoring systems continuously verify channel power and presence to prevent spectrum tilt or missing wavelengths that could destabilize Raman gain.

In SDM, these operational demands multiply. Splicing and connectorization of MCFs require micron-level precision to align all cores; misalignment can cause loss or inter-core leakage. Specialized fusion splicers and multi-core connectors are under development, employing MT-type ferrules or custom alignment ferrules. Maintenance also becomes more complex—damage to one MCF can impact multiple spatial channels simultaneously, necessitating new protection and restoration strategies.

Component availability remains limited. While commercial AWGs and EDFAs are abundant for WDM, multi-core or multi-mode versions are largely in research stages. Prototype multi-core EDFAs using shared-pump schemes have shown promising performance but are not yet mass-produced. As a result, SDM deployment currently requires co-development of the entire ecosystem, including amplifiers, connectors, transceivers, and monitoring tools.

Outlook and ongoing research

Despite these challenges, the optical community has repeatedly demonstrated its ability to overcome perceived limits. WDM, once doubted as overly complex, now forms the backbone of every long-haul network. Similarly, SDM has evolved from concept to experimental reality, achieving petabit-per-second transmission in laboratory settings.

Ongoing initiatives such as NICT’s EXAT and the EU’s MCF2020 projects are translating SDM technology into field trials—testing multi-core fibers in urban ducts and subsea environments. Early commercial applications are expected in short-haul data-center interconnects, where a few-core fiber can replace large cable bundles, and in submarine cables, where shared amplification across multiple cores can reduce repeater count and power consumption.

Ultimately, SDM and advanced multiplexing mark a paradigm shift in optical networking — extending scaling into the spatial dimension. While cost, manufacturability, and interoperability remain challenges, the rapid pace of progress suggests that space-multiplexed optical systems will soon transition from laboratory experiments to practical high-capacity infrastructure.

Canadian-engineered WDM optics by Technologie Optic Inc

At Technologie Optic Inc., we transform the theory of Wavelength Division Multiplexing into practical, high-performance connectivity. Our product portfolio spans the full spectrum of optical transmission speeds—from 1 G to 800 G—supporting both CWDM and DWDM wavelength grids across the O, E, S, C, and L bands. Each transceiver is meticulously engineered for compatibility with leading network platforms, ensuring seamless deployment in data-center, metro, and long-haul environments.

Our DWDM portfolio covers over 70 standardized ITU channels (1520–1577 nm), enabling dense, high-capacity optical links that align perfectly with modern EDFA- and ROADM-based infrastructures. Complemented by advanced CWDM modules, our solutions provide scalable, cost-efficient options for both access and backbone networks.

By combining rigorous R&D with in-house optical testing and precise wavelength control, Optic.ca guarantees transceiver stability, low insertion loss, and full digital diagnostics (DDM). This commitment to quality and interoperability makes Optic.ca a trusted Canadian manufacturer in the evolution toward multi-terabit WDM and future SDM optical systems—bridging research innovation with deployable telecom excellence.

Conclusion

Wavelength Division Multiplexing has been the foundation of optical capacity growth for more than two decades. By allowing multiple wavelength channels to coexist on a single fiber, it maximizes spectral efficiency and has supported the exponential rise in global data traffic without constant new fiber deployment. This study reviewed the principles of CWDM and DWDM, as well as the enabling filter technologies such as thin-film filters, arrayed waveguide gratings, and fiber Bragg gratings. Modern systems primarily operate within the low-loss C and L bands, with research expanding into the S band and beyond to exploit the full optical spectrum. However, as networks approach the ~100 Tb/s capacity limit of single-mode fibers, new paradigms are required to overcome this “capacity crunch.”

Space-Division Multiplexing offers the most promising path forward. By transmitting multiple spatial channels—using multi-core or few-mode fibers—SDM multiplies capacity without increasing per-channel bandwidth. Laboratory demonstrations have already achieved petabit-per-second throughput in single fibers, proving the scalability of this approach. Yet practical deployment faces hurdles, including inter-core crosstalk, mode coupling, and the cost of specialized components such as multi-core amplifiers and fan-in/fan-out couplers. The development of standardized geometries and connector interfaces will be essential for interoperability and large-scale adoption. The likely near-term evolution will combine WDM and SDM, with a few cores—each carrying dozens of DWDM channels—forming hybrid high-capacity links. Alongside multi-band operation (C+L+S) and advanced modulation formats, such systems could push total capacities into the hundreds of terabits or even petabits per second. In summary, WDM will remain the cornerstone of optical communications, while SDM introduces a new spatial dimension of scalability. Together, they ensure that optical fibers—through innovation in amplification, modulation, and processing—will continue to meet the world’s ever-growing demand for bandwidth well into the future.