WDM Fronthaul vs Passive Fronthaul

Fronthaul is the most demanding segment of the 5G transport network. It connects the radio unit at the antenna site to the distributed unit that performs baseband processing, carrying digitized radio samples or partially processed baseband data. The bandwidth is high, the latency budget is measured in tens of microseconds, and the synchronization must be maintained to sub-microsecond accuracy. How that traffic is multiplexed onto fiber determines the capacity, reach, cost, and operational complexity of the entire radio access network.

Two multiplexing approaches dominate fronthaul deployment. WDM fronthaul assigns a separate wavelength to each fronthaul link, combining many wavelengths onto a shared fiber pair using wavelength division multiplexing. Passive fronthaul shares a single wavelength across multiple links using optical power splitters, dividing the optical power among the connected endpoints in a point-to-multipoint topology. The two approaches are sometimes confused because both use passive optical components and both reduce the fiber count compared to dedicated point-to-point links. But they work on fundamentally different principles—wavelength separation versus power division—and the differences have profound implications for capacity, reach, and deployment topology.

The distinction matters because it determines how the fronthaul network scales. A WDM system can add capacity by adding wavelengths, without changing the fiber plant or the power budget of existing links. A passive splitter system divides a fixed optical power budget among all connected endpoints, so adding endpoints reduces the power available to each and may require higher-power transmitters or shorter reach. Understanding when each approach is the right choice—and where they can be combined—is essential for designing fronthaul networks that can support the densification of 5G and the migration to C-RAN architectures.

This guide examines both approaches in depth: their architecture, multiplexing principles, bandwidth and power budgets, latency and synchronization characteristics, fiber consumption, cost structures, operational complexity, and the deployment 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. 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 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. In fronthaul deployments, passive WDM is often preferred because it introduces no jitter, requires no power, and is transparent to the synchronization protocol.

2.1 Characteristics of WDM Fronthaul

• Wavelength assignment: Each fronthaul link has its own dedicated wavelength.

• Dedicated bandwidth: Each wavelength carries the full bandwidth of its assigned link, independent of the other wavelengths.

• Point-to-point per wavelength: Each wavelength establishes a dedicated logical connection between one RU and one DU port.

• Passive multiplexing: The mux/demux is a passive optical component with no power and no active processing.

• Scalable by wavelength: Additional links are added by adding wavelengths, without changing the fiber plant.

• Fiber efficiency: A single fiber pair can carry 8 to 40+ fronthaul links, depending on the WDM technology.

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.

3. What Is Passive Fronthaul?

Passive fronthaul uses a single wavelength shared across multiple links through optical power splitters. Instead of assigning a separate wavelength to each link, the passive fronthaul system operates on one wavelength and divides the optical power among the connected endpoints in a point-to-multipoint topology.

The most common implementation of passive fronthaul is based on Passive Optical Network (PON) technology. A single optical line terminal port at the baseband hotel transmits downstream on one wavelength, and a passive optical splitter divides the signal among multiple radio units. Upstream, the radio units transmit on a different wavelength, and the splitter combines their signals back to the line terminal. Because all radio units share the same upstream wavelength, a multiple access protocol—typically time division multiple access (TDMA)—is required to arbitrate upstream transmission.

3.1 Characteristics of Passive Fronthaul

• Shared wavelength: All links share the same wavelength, separated by time rather than by wavelength.

• Power splitting: The optical splitter divides the available optical power among the connected endpoints.

• Point-to-multipoint: A single port at the baseband hotel serves multiple radio units through the splitter.

• Shared bandwidth: The total bandwidth of the wavelength is shared among all connected endpoints.

• Fiber efficiency: A single fiber can serve multiple radio units through the splitter, reducing the fiber count.

• Protocol dependency: The multiple access protocol (TDMA, FDM, or other) determines the latency and bandwidth allocation.

Passive fronthaul is attractive in access networks where the fiber to each radio site is a single strand, or where the splitter can be placed close to a cluster of radio units. It is the architecture used in PON-based mobile fronthaul and in some small cell deployments.

4. Wavelength Assignment vs Power Splitting

The fundamental difference between WDM fronthaul and passive fronthaul is how the available optical resource is divided.

WDM divides the optical spectrum. Each fronthaul link is assigned a unique wavelength, and the wavelengths are combined onto a shared fiber by a passive multiplexer. Because each link has its own wavelength, it carries its own data independently of the other links. There is no contention, no shared bandwidth, and no multiple access protocol. The multiplexer is a wavelength-selective device: it combines and separates signals by wavelength, not by power.

Passive fronthaul divides the optical power. All links share the same wavelength, and the splitter divides the optical power among them. Because the links share the same wavelength, they must share the same bandwidth, and a multiple access protocol is required to arbitrate transmission. The splitter is a power-splitting device: it divides the optical signal among multiple output fibers, reducing the power available to each.

The choice between them is not merely technical. It determines whether each radio unit has dedicated bandwidth or shares bandwidth with its neighbors, whether the fronthaul latency is deterministic or depends on the multiple access protocol, and whether the power budget can be expanded by adding wavelengths or must be managed by limiting the split ratio or increasing transmitter power.

5. Bandwidth and Capacity

Bandwidth is where the two approaches diverge most sharply.

5.1 WDM Fronthaul Bandwidth

In WDM fronthaul, each link has a dedicated wavelength carrying the full bandwidth of the transceiver. A 25G WDM transceiver carries 25 Gbps on its wavelength, independent of the other wavelengths on the same fiber. Adding a new link means adding a new wavelength, which adds 25 Gbps of capacity without affecting the existing links.

This dedicated bandwidth model is well-suited to fronthaul, where each radio unit requires a fixed, guaranteed bandwidth. The 25 Gbps of a 25G eCPRI link is the bandwidth required by the radio unit; it is not shared with any other link, and its availability does not depend on the traffic patterns of neighboring radio units.

5.2 Passive Fronthaul Bandwidth

In passive fronthaul, all links share the same wavelength and therefore the same total bandwidth. If the shared wavelength operates at 25 Gbps and four radio units are connected, each radio unit has an average of 6.25 Gbps—but the actual allocation depends on the multiple access protocol and the instantaneous traffic demand.

Time division multiple access allocates the upstream bandwidth in time slots. Each radio unit transmits in its assigned time slots, and the total bandwidth is divided among the units. The downstream bandwidth is typically broadcast or time-multiplexed as well. The result is that each radio unit's effective bandwidth is a fraction of the total, and the fraction depends on the number of units and the allocation algorithm.

For fronthaul, this shared bandwidth model is problematic. Fronthaul traffic is not bursty in the way that residential broadband traffic is; it is a continuous, high-bandwidth stream of digitized radio samples. Sharing a single wavelength among multiple radio units means that each unit must either reduce its bandwidth (which degrades radio performance) or contend for time slots (which adds latency and jitter). The shared bandwidth model of passive fronthaul is therefore better suited to lower-bandwidth applications—such as small cell backhaul or residential PON—than to high-bandwidth fronthaul.

6. Latency and Synchronization

Latency and synchronization are the two performance parameters that most directly affect fronthaul viability.

6.1 Latency in WDM Fronthaul

WDM 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 passive multiplexer and demultiplexer add negligible latency—typically in the nanosecond range—because they do not process the signal. The result is deterministic latency that depends only on the fiber length and the transceiver propagation delay.

6.2 Latency in Passive Fronthaul

Passive fronthaul latency is also dominated by fiber propagation, but the multiple access protocol introduces additional latency and jitter. In a TDMA-based passive fronthaul system, each radio unit must wait for its assigned time slots before transmitting upstream. The waiting time depends on the number of units and the allocation algorithm, and it can be significant—tens of microseconds or more. This additional latency is not present in WDM fronthaul, where each link has its own wavelength and can transmit continuously.

The jitter introduced by the multiple access protocol is also a concern. Fronthaul synchronization requires tight phase alignment between the RU and the DU, and any jitter in the transport path can degrade the synchronization accuracy. WDM fronthaul, being point-to-point per wavelength, introduces no protocol-related jitter. Passive fronthaul, with its shared upstream and time-slot arbitration, may introduce jitter that must be compensated by the synchronization system.

6.3 Synchronization Transparency

Both WDM and passive fronthaul use passive optical components, so both are transparent to the synchronization protocol. SyncE and IEEE 1588 PTP pass through the passive multiplexer or splitter without degradation. The difference is in the protocol-related jitter: WDM fronthaul carries the synchronization information on a dedicated wavelength with no contention, while passive fronthaul shares the upstream with data traffic and may be affected by the time-slot arbitration.

7. Power Budget

Power budget is where passive fronthaul faces its most fundamental constraint.

7.1 WDM Fronthaul Power Budget

In WDM fronthaul, each wavelength passes through the multiplexer and demultiplexer with a fixed insertion loss—typically 2 to 4 dB per mux/demux pair. The power budget for each wavelength is independent of the other wavelengths: adding a new wavelength does not reduce the power available to existing wavelengths. This is because the multiplexer combines and separates wavelengths without dividing power among them.

The WDM power budget is therefore scalable. Adding a new link adds a new wavelength with its own power budget, and the existing links are unaffected. The only limit is the channel count of the multiplexer and the total optical power that the fiber can carry before nonlinear effects become significant.

7.2 Passive Fronthaul Power Budget

In passive fronthaul, the splitter divides the optical power among all connected endpoints. A 1:2 splitter divides the power in half, introducing 3 dB of loss. A 1:4 splitter introduces 6 dB of loss. A 1:8 splitter introduces 9 dB, a 1:16 splitter introduces 12 dB, and a 1:32 splitter introduces 15 dB. These losses are fundamental to the power-splitting principle and cannot be avoided.

The power budget for each endpoint is therefore reduced as the split ratio increases. Adding a new endpoint to the splitter reduces the power available to all existing endpoints, not just the new one. This is a fundamental difference from WDM: in WDM, adding a wavelength does not affect the existing wavelengths; in passive fronthaul, adding an endpoint reduces the power available to every endpoint.

The power budget constraint limits the split ratio and the reach of passive fronthaul. A 1:16 splitter with 12 dB of loss leaves little margin for fiber loss, connector loss, and aging, especially at 25G where the transceiver power budget is tighter than at lower speeds. In practice, passive fronthaul at 25G is often limited to 1:8 or 1:16 splits over short reaches, while lower-speed applications can support higher split ratios.

8. Fiber Consumption

Both WDM and passive fronthaul reduce the fiber count compared to dedicated point-to-point links, but they do so in different ways.

8.1 WDM Fronthaul Fiber Consumption

WDM fronthaul uses one fiber pair for the shared trunk between the baseband hotel and the remote node, plus one fiber pair per radio site for the drop from the remote node to the site. The trunk carries all the wavelengths; the drops carry the individual wavelengths.

In a typical C-RAN deployment, the trunk runs from the baseband hotel to a remote optical node near the cluster of radio sites. From the remote node, individual fiber pairs run to each radio site. The trunk carries the WDM signal; the drops carry the individual wavelengths.

8.2 Passive Fronthaul Fiber Consumption

Passive fronthaul uses one fiber for the shared trunk and one fiber for each drop from the splitter to the radio unit. The splitter is placed at a point where the drops can fan out to the radio sites.

The fiber consumption is similar to WDM: one trunk fiber (or fiber pair) and one drop per site. The difference is that in WDM, the trunk carries multiple wavelengths (one per site), while in passive fronthaul, the trunk carries a single wavelength shared by all sites.

The fiber consumption is similar, but the capacity is not. A WDM trunk carrying 16 wavelengths provides 16 × 25 Gbps = 400 Gbps of total capacity, with each site receiving its full 25 Gbps. A passive fronthaul trunk carrying one shared wavelength provides 25 Gbps of total capacity, shared among all 16 sites, giving each site an average of 1.56 Gbps—far below what a 5G radio unit requires.

9. Cost Comparison

The cost comparison between WDM and passive fronthaul depends on the number of links, the split ratio, and the bandwidth requirement.

9.1 WDM Fronthaul Cost

WDM fronthaul cost includes the WDM transceivers, the passive multiplexer and demultiplexer, and the fiber. The transceivers are the dominant cost: a 25G WDM transceiver costs more than a standard 25G grey transceiver because it operates on a specific wavelength. The multiplexer and demultiplexer are passive and relatively inexpensive. The fiber cost is shared across all wavelengths, so the per-link fiber cost is low.

9.2 Passive Fronthaul Cost

Passive fronthaul cost includes the PON transceivers (which are shared at the line terminal and individual at each radio unit), the passive splitter, and the fiber. The splitter is inexpensive—a passive 1:16 splitter costs far less than a 16-channel WDM multiplexer. The transceivers are also less expensive than WDM transceivers because they operate on a single wavelength rather than a specific wavelength in a WDM plan. But the shared bandwidth model means that passive fronthaul is only suitable for lower-bandwidth applications, so the cost comparison must account for the bandwidth limitation.

The cost comparison is not straightforward because the two approaches serve different bandwidth requirements. WDM fronthaul provides dedicated 25 Gbps per link, which is what 5G fronthaul requires. Passive fronthaul provides shared bandwidth, which is sufficient for lower-speed applications but not for high-bandwidth 5G fronthaul. The cost comparison must therefore be made on a per-gigabit basis, and WDM generally provides a lower cost per gigabit for high-bandwidth fronthaul.

10. Operational Complexity

Operational complexity differs between the two approaches and affects the total cost of ownership.

10.1 WDM Fronthaul Operational Characteristics

• Wavelength planning: Each link must be assigned a wavelength, and the plan must be managed across the network.

• Wavelength-specific transceivers: Each link requires a transceiver tuned to its assigned wavelength, complicating sparing.

• Multiplexer management: The multiplexer and demultiplexer must be installed, configured, and maintained.

• Wavelength-aware test equipment: Testing requires a wavelength-aware power meter or optical spectrum analyzer.

• Deterministic behavior: Once deployed, each wavelength behaves predictably, with no protocol-related variability.

10.2 Passive Fronthaul Operational Characteristics

• No wavelength planning: All endpoints share the same wavelength, so there is no wavelength plan to manage.

• Standardized transceivers: All endpoints use the same transceiver type, simplifying sparing.

• Splitter management: The splitter is a passive component with no configuration, but its split ratio determines the power budget for all endpoints.

• Protocol management: The multiple access protocol must be configured and maintained, and it may require tuning as the number of endpoints changes.

• Power budget management: Adding an endpoint reduces the power available to all endpoints, which may require re-engineering the link budget.

WDM fronthaul is more complex in wavelength management and transceiver sparing, but it is deterministic and predictable once deployed. Passive fronthaul is simpler in wavelength management but requires protocol management and has a power budget that is coupled across all endpoints.

11. Deployment Scenarios

The choice between WDM and passive fronthaul depends on the deployment scenario. Several common scenarios are summarized below.

12. Comparison Summary

13. Selection Framework

14. Emerging Trends

14.1 WDM-PON for Fronthaul

WDM-PON combines wavelength division multiplexing with the point-to-multipoint topology of PON. Each endpoint is assigned its own wavelength, but the wavelengths are distributed through a passive wavelength router rather than a power splitter. This provides dedicated bandwidth per endpoint with the fiber savings of a point-to-multipoint topology. WDM-PON is emerging as a promising approach for fronthaul because it combines the dedicated bandwidth of WDM with the fiber efficiency of PON.

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 favors WDM, because passive fronthaul's shared bandwidth model becomes even more limiting at higher speeds.

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. Passive fronthaul is less suited to this convergence because its shared wavelength model cannot differentiate between traffic classes.

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

WDM fronthaul and passive fronthaul are two approaches to multiplexing fronthaul traffic onto fiber. WDM assigns each link its own wavelength, providing dedicated bandwidth, deterministic latency, and scalability by wavelength. Passive fronthaul shares a single wavelength across multiple links through optical power splitters, providing fiber savings and simplified transceiver management at the cost of shared bandwidth and reduced power budget per endpoint.

The choice between them is governed primarily by the bandwidth requirement. 5G fronthaul requires dedicated high bandwidth per radio unit—typically 25 Gbps for eCPRI—which WDM provides and passive fronthaul cannot. Passive fronthaul is better suited to lower-bandwidth applications such as small cell backhaul, residential PON, or compressed CPRI fronthaul, where the shared bandwidth model is sufficient.

Latency and synchronization also favor WDM for fronthaul. WDM provides deterministic latency with no protocol-related jitter, while passive fronthaul introduces time-slot arbitration delay and protocol-dependent jitter. For fronthaul, where the latency budget is measured in tens of microseconds and synchronization must be maintained to sub-microsecond accuracy, this difference is significant.

The power budget is another decisive factor. WDM adds a fixed insertion loss per wavelength, and adding a wavelength does not affect the power available to existing wavelengths. Passive fronthaul divides the optical power among all endpoints, so adding an endpoint reduces the power available to every endpoint. This coupling limits the split ratio and the reach of passive fronthaul, especially at 25G where the power budget is tighter.

As 5G networks densify and fronthaul bandwidth requirements increase, WDM fronthaul will remain the primary approach for high-bandwidth 5G transport. Passive fronthaul will continue to serve lower-bandwidth applications where its fiber savings and simplified management are advantageous. The two approaches are complementary, and the choice between them should be based on the bandwidth, latency, and synchronization requirements of the specific deployment.

16. Q&A