TEL：+86 132 6656 7067 
Fronthaul is the transport segment that connects a 5G radio unit to its distributed unit. It is the most demanding part of the 5G transport network: the bandwidth requirement is high, the latency budget is measured in tens of microseconds, and the synchronization accuracy must be maintained to sub-microsecond levels. Two architectural approaches dominate fronthaul deployment: dark fiber and WDM.
Dark fiber fronthaul uses unlit fiber pairs—fiber that is leased or owned but not equipped with active optical transmission equipment. Each radio unit or radio site gets its own dedicated fiber pair, and the optical signal between the radio and the baseband processing travels directly over that fiber without wavelength multiplexing. It is the simplest approach: point-to-point fiber, a transceiver at each end, and nothing in between. Its limitation is fiber consumption—one fiber pair per fronthaul link, which becomes expensive when many radio sites must be connected.
WDM fronthaul uses wavelength division multiplexing to combine multiple fronthaul links onto a single fiber pair. Instead of one fiber pair per radio site, a WDM system assigns a different wavelength to each site or each link, multiplexes them onto a shared fiber, and demultiplexes them at the far end. A single fiber pair can carry 8, 16, or 40 fronthaul links, depending on the WDM technology and the wavelength plan. The trade-off is additional optical equipment—multiplexers, demultiplexers, amplifiers, and possibly wavelength converters—at each end of the shared fiber.
The choice between them depends on three factors: the number of radio sites to be connected, the cost and availability of fiber, and the operational complexity the operator is prepared to manage. In dense urban areas where fiber is expensive or scarce, WDM is often the only economically viable approach. In rural or suburban areas where dark fiber is available and radio site density is low, dark fiber may be simpler and cheaper. In many real deployments, the two approaches are combined: dark fiber for the first few sites and WDM as the fiber plant reaches its capacity limit.
This guide examines both approaches in depth: their architecture, fiber consumption, latency, synchronization, capacity scaling, cost, operational complexity, and the scenarios where each is the right choice.
1. What Is 5G Fronthaul?
Fronthaul is the transport link between the Radio Unit (RU) and the Distributed Unit (DU) in a disaggregated 5G RAN. The RU sits at the antenna site and performs radio frequency processing and analog-to-digital conversion. The DU performs the upper physical layer and real-time baseband processing. Between them, the fronthaul carries digitized radio samples or partially processed baseband data.
The functional split point between the RU and the DU determines the fronthaul bandwidth and latency requirements. The most demanding split—option 8, corresponding to traditional CPRI—places the split between the physical layer and the radio frequency, requiring the fronthaul to carry raw digitized radio samples. The bandwidth for a single 100 MHz 5G sector can reach 25 Gbps, and the one-way latency budget is less than 100 microseconds.
A less demanding split—option 7-2, corresponding to eCPRI—places the split within the physical layer, reducing the fronthaul bandwidth to approximately 10 to 25 Gbps per sector and relaxing the latency budget slightly. The most relaxed split—option 2, corresponding to the midhaul interface—places the split between the radio link control and packet data convergence protocol layers, reducing the bandwidth to 100 Mbps to 1 Gbps and relaxing the latency budget to 1 to 10 milliseconds.
Regardless of the split point, the fronthaul must deliver high bandwidth, low latency, and precise synchronization. It must also be highly reliable, because a fronthaul failure takes the radio site offline entirely.
2. What Is Dark Fiber Fronthaul?
Dark fiber fronthaul uses a dedicated fiber pair between the RU and the DU. The fiber may be owned by the operator or leased from a fiber provider. It is "dark" because it carries no other traffic and no wavelength multiplexing—it is used exclusively for the fronthaul link between one RU and one DU.
The optical signal between the RU and the DU is typically carried by a single transceiver at each end, using a single wavelength. For short reaches—up to a few kilometers—direct-detect optics at 10G or 25G are sufficient. For longer reaches, higher-power optics or coherent transmission may be required.
2.1 Characteristics of Dark Fiber Fronthaul
Dedicated fiber pair per link: One fiber pair per RU-to-DU connection. No sharing, no multiplexing.
Simple architecture: Two transceivers and a fiber pair. No multiplexers, no amplifiers, no wavelength planning.
Deterministic latency: The fiber path is fixed and direct, so the latency is determined solely by the fiber length and the transceiver propagation delay.
Full control: The operator controls the entire path, including the fiber, the transceivers, and the synchronization chain.
Fiber consumption: One fiber pair per link, which becomes expensive as the number of links grows.
Dark fiber is the simplest and most direct approach to fronthaul. It is also the most fiber-hungry. A C-RAN cluster with 20 radio sites requires 20 fiber pairs running from each site to the baseband hotel. In a dense urban environment where each fiber pair must be leased or installed, this can be prohibitively expensive.
3. What Is WDM Fronthaul?
WDM fronthaul uses wavelength division multiplexing to combine multiple fronthaul links onto a single fiber pair. Each link or each radio site is assigned a different wavelength, and a passive or active multiplexer combines the wavelengths onto the shared fiber. At the far end, a demultiplexer separates the wavelengths and routes each to its corresponding DU port.
The WDM system can be passive—using only passive multiplexers and demultiplexers, with no powered equipment in the path—or active—using optical amplifiers, wavelength converters, or reconfigurable optical add/drop multiplexers to manage the wavelengths dynamically.
3.1 Characteristics of WDM Fronthaul
Wavelength sharing: Multiple fronthaul links share a single fiber pair, each on its own wavelength.
Fiber efficiency: A single fiber pair can carry 8, 16, or 40 fronthaul links, depending on the WDM technology.
Additional equipment: Multiplexers, demultiplexers, and possibly amplifiers are required at each end of the shared fiber.
Wavelength planning: Each link must be assigned a wavelength, and the plan must be managed across the network.
Scalability: Additional links can be added by adding wavelengths, without pulling new fiber.
Latency: The WDM equipment introduces a small amount of additional latency, typically in the nanosecond to low microsecond range.
WDM fronthaul is the standard approach for C-RAN deployments where many radio sites must be connected to a centralized baseband hotel. It reduces the fiber count from one pair per site to a shared pair, which is essential in dense urban environments where fiber is scarce or expensive.
4. CWDM vs DWDM for Fronthaul
Two WDM technologies are used in fronthaul deployments: Coarse Wavelength Division Multiplexing (CWDM) and Dense Wavelength Division Multiplexing (DWDM).
4.1 CWDM
CWDM uses wider wavelength spacing—typically 20 nm—across a broader wavelength range, from 1271 nm to 1611 nm. It supports up to 18 channels, though 8-channel and 16-channel configurations are most common in fronthaul. CWDM multiplexers are passive and relatively inexpensive, and the wider wavelength spacing allows the use of uncooled lasers, reducing module cost and power consumption.
CWDM is well-suited to fronthaul because it provides enough channel count for most C-RAN clusters, its passive multiplexers are simple and reliable, and its uncooled lasers keep the cost and power of the fronthaul optics low.
4.2 DWDM
DWDM uses narrower wavelength spacing—typically 0.8 nm or 0.4 nm—in the C-band (1530–1565 nm), supporting 40, 80, or more channels. DWDM allows much higher channel counts on a single fiber pair, but requires temperature-controlled lasers and more precise multiplexers, increasing cost and power consumption.
DWDM is used in fronthaul when the number of radio sites exceeds what CWDM can support, or when the shared fiber is extremely scarce and the operator needs to maximize the capacity of each fiber pair. It is more common in metro and long-haul transport than in fronthaul, but it is deployed in dense urban C-RAN clusters where fiber is at a premium.
| Characteristic | CWDM | DWDM |
|---|---|---|
| Wavelength Spacing | 20 nm | 0.8 nm or 0.4 nm |
| Wavelength Range | 1271–1611 nm | C-band (1530–1565 nm) |
| Max Channels | 18 | 40–80+ |
| Laser Type | Uncooled | Temperature-controlled |
| Mux/Demux Cost | Low | Higher |
| Optics Cost | Lower | Higher |
| Power Consumption | Lower | Higher |
| Typical Fronthaul Use | 8–16 channels | 40+ channels |
5. Fiber Consumption
Fiber consumption is the primary differentiator between the two approaches. Dark fiber consumes one fiber pair per fronthaul link. WDM consumes one fiber pair for the entire group of links, plus the wavelengths assigned to each link.
| Number of Fronthaul Links | Dark Fiber (Fiber Pairs) | CWDM 8-Channel (Fiber Pairs) | DWDM 40-Channel (Fiber Pairs) |
|---|---|---|---|
| 1 | 1 | 1 | 1 |
| 8 | 8 | 1 | 1 |
| 16 | 16 | 2 | 1 |
| 40 | 40 | 5 | 1 |
| 80 | 80 | 10 | 2 |
The fiber savings of WDM are dramatic. For 16 fronthaul links, dark fiber requires 16 fiber pairs; an 8-channel CWDM system requires only 2 fiber pairs; a DWDM system requires only 1. In a dense urban environment where each fiber pair costs thousands of dollars to lease or install, these savings can be decisive.
The savings are also important at the baseband hotel. In a dark fiber deployment, the baseband hotel must terminate one fiber pair per radio site, requiring a patch panel and transceiver port for each. In a WDM deployment, the baseband hotel terminates only the shared fiber pair, and the WDM demultiplexer routes each wavelength to the appropriate DU port. This reduces the patch panel size, the transceiver count, and the rack space required at the hotel.
6. Latency and Synchronization
Latency and synchronization are the two performance parameters that most directly affect fronthaul viability.
6.1 Latency
Fronthaul latency is dominated by the fiber propagation delay. Light travels through fiber at approximately 5 microseconds per kilometer, so a 10-kilometer fronthaul link introduces 50 microseconds of one-way propagation delay. The one-way latency budget for eCPRI fronthaul is typically less than 100 microseconds, so the fiber length is a primary constraint.
Both dark fiber and WDM fronthaul are subject to the same fiber propagation delay. The difference is the additional latency introduced by the WDM equipment—multiplexers, demultiplexers, and amplifiers—which is typically in the nanosecond to low microsecond range. This additional latency is small compared to the fiber propagation delay and is generally not a deciding factor between the two approaches.
6.2 Synchronization
Fronthaul requires frequency and phase synchronization between the RU and the DU. Frequency accuracy must be maintained to 50 parts per billion or better, and phase accuracy must be maintained to approximately 1.5 microseconds for basic operation and sub-microsecond for advanced features like coordinated multipoint and massive MIMO.
Both dark fiber and WDM fronthaul can carry synchronization information. In dark fiber, the synchronization is carried directly on the fiber via SyncE or IEEE 1588 Precision Time Protocol. In WDM, the synchronization must be carried on one of the wavelengths—typically a dedicated synchronization channel or one of the fronthaul channels—and the WDM equipment must be transparent to the synchronization protocol.
Passive WDM equipment is inherently transparent to synchronization, because it does not process the optical signal. Active WDM equipment—amplifiers, wavelength converters, and ROADMs—may introduce jitter and wander that affect synchronization accuracy. This is one reason why passive WDM is preferred for fronthaul: it preserves the synchronization characteristics of the direct fiber path.
| Parameter | Dark Fiber | WDM Fronthaul |
|---|---|---|
| Fiber Propagation Delay | ~5 µs/km | ~5 µs/km |
| Equipment Latency | Negligible | Nanoseconds to low microseconds |
| Synchronization Transparency | Native | Passive: transparent; Active: may add jitter |
| Typical Fronthaul Latency | Fiber length dependent | Fiber length + equipment delay |
7. Capacity Scaling
Capacity scaling is where WDM's advantage becomes most apparent as the network grows.
7.1 Dark Fiber Scaling
Dark fiber scales linearly. Each new radio site requires a new fiber pair from the site to the baseband hotel. In a C-RAN cluster with 40 sites, 40 fiber pairs must be provisioned, terminated, and managed. The fiber plant at the baseband hotel must accommodate 40 patch panel ports and 40 transceivers, and the cable trays must carry 40 fiber pairs.
The linear scaling of dark fiber means that the cost of fiber grows proportionally with the number of sites. In a dense urban environment, this can quickly become the dominant cost of the fronthaul deployment.
7.2 WDM Scaling
WDM scales by adding wavelengths, not fibers. When a new radio site is added, it is assigned a new wavelength and connected to the existing shared fiber pair. No new fiber is pulled, no new patch panel port is consumed at the baseband hotel, and no new cable tray capacity is required. The only additions are the WDM transceiver at the new site and the corresponding port on the WDM multiplexer.
The WDM system can be scaled incrementally. An 8-channel CWDM system can be upgraded to 16 channels by replacing the multiplexer with a higher-channel-count unit, without changing the fiber plant. This incremental scalability makes WDM particularly attractive in deployments where the number of radio sites is expected to grow over time.
8. Power and Optical Budget
Both approaches must satisfy an optical power budget, but the budget is more complex for WDM because of the additional passive components in the path.
8.1 Dark Fiber Power Budget
The dark fiber power budget is simple: the transmitter output power must exceed the fiber loss plus the connector losses plus the receiver sensitivity. For a 10-kilometer single-mode link with 0.3 dB/km fiber loss, the fiber loss is 3 dB. Adding 2 dB for connectors and splices, the total loss is 5 dB. A 25G LR transceiver with 6 dB of budget can span this link with margin.
8.2 WDM Power Budget
The WDM power budget includes the fiber loss, the connector losses, the multiplexer and demultiplexer insertion loss, and any amplifier gain or loss. A CWDM multiplexer typically introduces 2 to 4 dB of insertion loss, and the demultiplexer introduces a similar amount. This additional loss must be accounted for in the link budget.
For short reaches, the additional WDM loss can be accommodated by using higher-power transceivers or by limiting the reach. For longer reaches, optical amplifiers may be required, which add cost, power consumption, and complexity. The amplifiers must also be compatible with the fronthaul wavelengths and must not introduce jitter that affects synchronization.
| Budget Element | Dark Fiber | WDM Fronthaul |
|---|---|---|
| Fiber Loss | 0.3–0.4 dB/km | 0.3–0.4 dB/km |
| Connector Loss | 0.2–0.5 dB per mated pair | 0.2–0.5 dB per mated pair (more connectors) |
| Mux/Demux Loss | None | 2–4 dB per mux/demux pair (CWDM) |
| Amplifier | Not required for short reach | Required for long reach |
| Typical 10 km Budget | ~5–7 dB | ~9–13 dB |
9. Cost Comparison
The cost comparison between dark fiber and WDM fronthaul depends heavily on the cost of fiber and the number of links.
9.1 Dark Fiber Cost
Dark fiber cost is dominated by the fiber itself. If the operator owns the fiber, the cost is the installation cost, amortized over the life of the fiber plant. If the operator leases the fiber, the cost is the recurring lease fee, typically per fiber pair per month or per year.
The equipment cost for dark fiber is low: one transceiver per end per link. For 20 links, that is 40 transceivers, plus the patch panels and cable management at each end.
9.2 WDM Cost
WDM cost is dominated by the optical equipment: the multiplexers and demultiplexers, the amplifiers (if required), and the WDM transceivers. The fiber cost is shared across many links, so the per-link fiber cost is much lower than dark fiber.
The trade-off is that WDM has higher equipment cost but lower fiber cost. The break-even point depends on the cost of fiber. In an environment where fiber is cheap and abundant, dark fiber may be the lower-cost option even at moderate link counts. In an environment where fiber is expensive or scarce, WDM becomes cost-effective at a relatively low link count—often 4 to 8 links.
| Cost Element | Dark Fiber (20 links) | CWDM 8-Channel (20 links) |
|---|---|---|
| Fiber Pairs Required | 20 | 3 (8+8+4) |
| Transceivers | 40 | 40 |
| Mux/Demux | None | 6 (3 pairs) |
| Amplifiers | None | None (short reach) |
| Patch Panel Ports (Hotel) | 20 | 3 |
| Fiber Lease Cost (Recurring) | 20 × rate | 3 × rate |
| Equipment Cost | Low | Moderate |
The comparison shows the trade-off clearly: WDM reduces the recurring fiber cost by a factor of 6 to 7, at the price of additional equipment cost. In a deployment where the fiber lease is the dominant cost, WDM pays for itself quickly. In a deployment where the operator owns the fiber and the incremental cost of using additional fiber pairs is low, dark fiber may be the more economical choice.
10. Operational Complexity
Operational complexity is often underestimated in the dark fiber vs WDM decision.
10.1 Dark Fiber Operational Characteristics
Simple provisioning: Connect a transceiver at each end and the link is up. No wavelength planning, no multiplexer configuration.
Simple troubleshooting: A link failure is either a fiber break or a transceiver failure. The diagnostic path is straightforward.
Simple sparing: A spare transceiver of the same type is all that is needed.
Linear growth: Each new site requires a new fiber pair and a new transceiver, with no interaction with existing links.
10.2 WDM Operational Characteristics
Wavelength planning: Each link must be assigned a wavelength, and the plan must be managed across the network to avoid conflicts.
Multiplexer management: The multiplexer and demultiplexer must be installed, configured, and maintained. Adding a new wavelength may require replacing the multiplexer if it is at capacity.
Complex troubleshooting: A link failure may be caused by the transceiver, the fiber, the multiplexer, the demultiplexer, or a wavelength conflict. The diagnostic path is more complex.
Sparing complexity: Spare transceivers must match the wavelengths in use, and spare multiplexers must match the channel count and configuration.
Incremental growth: Adding a new site may require a new wavelength assignment and a multiplexer upgrade, but no new fiber.
The operational complexity of WDM is manageable but real. It requires wavelength planning discipline, more sophisticated test equipment, and staff trained in WDM operations. In organizations where these capabilities exist, WDM is not a barrier. In organizations where they do not, dark fiber may be the more practical choice even when WDM is technically feasible.
11. Deployment Scenarios
The choice between dark fiber and WDM fronthaul depends on the deployment scenario. Several common scenarios are summarized below.
| Scenario | Recommended Approach | Rationale |
|---|---|---|
| Rural macro site, few sites, long distance | Dark fiber | Fiber available; WDM equipment cost not justified for low link count |
| Suburban cluster, 4–8 sites | Evaluate both | Break-even range; depends on fiber cost and lease terms |
| Dense urban C-RAN, 8+ sites | CWDM WDM | Fiber scarce or expensive; WDM reduces fiber count by 8× |
| Dense urban C-RAN, 40+ sites | DWDM WDM | CWDM channel count insufficient; DWDM provides 40+ channels |
| 5G small cell densification | WDM | Many sites, limited fiber; WDM is the only scalable option |
| Baseband hotel with limited rack space | WDM | WDM reduces patch panel ports and transceiver count at the hotel |
| Operator-owned fiber, low incremental cost | Dark fiber | Fiber is already paid for; no need for WDM equipment |
| Leased fiber, high recurring cost | WDM | WDM reduces recurring fiber cost by 6–8× |
12. Comparison Summary
| Dimension | Dark Fiber Fronthaul | WDM Fronthaul |
|---|---|---|
| Fiber Consumption | 1 fiber pair per link | 1 fiber pair shared by 8–40 links |
| Architecture | Point-to-point, transceivers only | Shared fiber with mux/demux |
| Latency | Fiber propagation only | Fiber propagation + equipment delay |
| Synchronization Transparency | Native | Passive WDM: transparent |
| Capacity Scaling | Linear (new fiber per link) | Wavelength addition (no new fiber) |
| Optical Budget | Simpler; no mux/demux loss | Additional 2–4 dB mux/demux loss |
| Equipment Cost | Low | Moderate to high |
| Fiber Cost | High (1 pair per link) | Low (shared pair) |
| Operational Complexity | Low | Moderate (wavelength planning) |
| Troubleshooting | Simple | More complex |
| Sparing | Simple | Wavelength-specific |
| Best For | Rural, low link count, owned fiber | Urban, high link count, leased fiber |
13. Selection Framework
| Evaluation Factor | Recommendation |
|---|---|
| Number of radio sites | 1–4: dark fiber. 8+: WDM. 4–8: evaluate both. |
| Fiber availability | Abundant, owned: dark fiber. Scarce, leased: WDM. |
| Fiber cost | Low incremental cost: dark fiber. High recurring cost: WDM. |
| Latency budget | Both are comparable; WDM adds negligible delay. |
| Synchronization requirements | Both can meet; passive WDM preserves transparency. |
| Baseband hotel space | Constrained: WDM reduces patch panel and transceiver count. |
| Operational capability | Limited WDM expertise: dark fiber. Strong WDM expertise: WDM. |
| Growth trajectory | Stable link count: dark fiber. Growing link count: WDM. |
| Channel count | ≤8: CWDM. 8–40: CWDM or DWDM. 40+: DWDM. |
| Capital budget | Low capital: dark fiber. Higher capital, lower OpEx: WDM. |
14. Emerging Trends
14.1 Passive WDM for Fronthaul
Passive WDM is becoming the preferred approach for fronthaul in urban C-RAN deployments. Passive multiplexers require no power, introduce no jitter, and have no active components to fail. They are transparent to synchronization and to the fronthaul protocol, making them compatible with any RU and DU vendor. The simplicity and reliability of passive WDM make it attractive for operators who want the fiber savings of WDM without the operational complexity of active optical equipment.
14.2 Higher-Speed Fronthaul
As 5G radio units move to higher bandwidth—100 MHz and 200 MHz carriers, massive MIMO, and higher-order modulation—the fronthaul bandwidth requirement is increasing. 25G is becoming the standard fronthaul interface, and 50G is emerging for the most demanding deployments. Higher-speed fronthaul requires WDM systems that can support higher per-wavelength data rates without increasing the channel spacing.
14.3 Integration with Midhaul and Backhaul
The convergence of fronthaul, midhaul, and backhaul onto a common transport infrastructure is a growing trend. In this model, a single WDM system carries fronthaul, midhaul, and backhaul traffic on different wavelengths, with the fronthaul wavelengths optimized for low latency and the backhaul wavelengths optimized for capacity. This convergence reduces the total fiber count and simplifies the transport architecture.
14.4 Tunable Optics for WDM Fronthaul
Tunable optics allow the wavelength of a transceiver to be set in software, eliminating the need to stock separate transceivers for each wavelength. In a WDM fronthaul deployment, tunable optics simplify sparing and provisioning: a single tunable transceiver can be configured to any wavelength in the WDM plan. Tunable optics are becoming more affordable and are increasingly used in WDM fronthaul deployments.
15. Conclusion
5G WDM fronthaul and dark fiber fronthaul are two approaches to connecting radio units to baseband processing. Dark fiber uses a dedicated fiber pair per link and is the simplest approach: two transceivers, a fiber pair, and nothing in between. WDM uses wavelength division multiplexing to share a single fiber pair across multiple links, reducing the fiber count by 8 to 40 times at the cost of additional optical equipment.
The choice between them depends primarily on the number of radio sites, the cost and availability of fiber, and the operational capabilities of the organization. In rural and suburban deployments with few sites and available dark fiber, dark fiber is simpler and often cheaper. In dense urban deployments with many sites and scarce fiber, WDM is the only scalable option. The break-even point typically falls between 4 and 8 links, depending on fiber cost and lease terms.
Latency and synchronization are comparable between the two approaches. Both are subject to the same fiber propagation delay, and passive WDM introduces negligible additional latency and preserves synchronization transparency. The primary performance difference is the optical budget: WDM adds 2 to 4 dB of multiplexer and demultiplexer loss, which must be accounted for in the link budget.
In many real deployments, the two approaches are combined. Dark fiber is used for the first few sites, and WDM is deployed as the fiber plant reaches its capacity limit. This hybrid approach captures the simplicity of dark fiber where it is sufficient and the fiber efficiency of WDM where it is needed. As 5G networks densify and fiber becomes an increasingly scarce resource, WDM fronthaul will play a growing role in the transport architecture.
>
>
>
>
>
>
>
>
