By Randy Eisenach / Nokia
Primarily used on subsea networks, noise loading is a method that fills unused channels with ASE (amplified spontaneous emission) noise, simulating a filled WDM system. Recently, there’s been some industry interest in using ASE noise loading on terrestrial networks in addition to traditional subsea applications. Historically, terrestrial WDM systems relied on embedded dynamic power management algorithms to ensure constant per-channel power levels and a flat gain response across the operating band. Both methods, active power management and ASE noise loading, work equally well, without significant performance differences between the two approaches.
WDM and ILA nodes actively monitor and adjust wavelength power levels and spectral “flatness” across the C-band or combined C+L bands to achieve optimal network performance. On terrestrial-based WDM networks, these adjustments are typically controlled by “dynamic power management” algorithms operating within each ROADM and ILA node.
Optical power management algorithms automatically maintain constant per-channel power levels as network changes occur due to wavelength additions or deletions. The power management algorithms make these adjustments by controlling optical amplifier power levels. Suppose a network is operating with (10) x 400G wavelengths. Suppose a network operating with (10) x 400G wavelengths. In that case, each transmitting at a target optical power of +3 dBm, adds (5) additional 4000G wavelengths the dynamic power management algorithms automatically increase each amplifier’s output power to ensure all channels remain at their + 3 dBm level target.
The dynamic power management algorithms constantly run in the background, monitoring per-channel power levels and making amplifier adjustments to maintain constant per-channel power levels and flat gain response as wavelengths are added or deleted from the network.
WDM node dynamic power management algorithms perform exceptionally well for terrestrial networks and have been the primary optical power control method for over 20 years.
Subsea Network Applications
Subsea systems typically use an alternative optical power management technique, primarily because subsea equipment resides under hundreds to thousands of meters of seawater. Repairing subsea cables or repeaters (ILAs) can cost over $1 million dollars for each repair. In addition, it can take weeks to dispatch a maintenance repair ship to a remote part of the ocean, haul the fiber cable from the ocean floor, and make any necessary repairs.
As a result of the high repair costs and long repair times, subsea “wet plant” equipment is designed with high levels of redundancy, along with the lowest possible failure rate (FIT). One way to lower the failure rate (FIT) is using simpler amplifier designs incorporating fewer components. The fewer the components in an amplifier, the fewer things can fail.
The dynamic power management algorithms used on terrestrial WDM networks are software or firmware control algorithms. However, these algorithms rely on additional hardware built into amplifiers for monitoring total and per-channel optical power levels, as well as for controlling and adjusting amplifier pump lasers and output power. These other hardware items can include optical power taps, optical channel monitors (OCM), control circuits, and microprocessors. On undersea repeaters (ILAs), all these additional components would slightly increase the unit’s failure rate. While the slight increase in FIT rate to support dynamic power management is insignificant on terrestrial nodes, where a failed card can easily be replaced, on repeater nodes that sit under 2,000 meters of water, the increase in FIT rates could lead to costly repair bills. As a result, subsea systems utilize ASE noise loading as a simpler, less complex alternative to “dynamic” power management algorithms to monitor and manage optical power levels.
ASE Noise Loading
Noise loading is a technique where optical amplified spontaneous emission (ASE) noise fills all unused channels, as shown in Figure 1. The network operates at total capacity with all channels occupied, either with actual traffic carrying wavelengths or ASE noise channels. As new channels are added to the network, the noise channels are replaced by the “live” traffic having wavelengths. Similarly, as “active” channels are deleted from the network, they are automatically replaced by ASE noise channels. To the WDM network and amplifiers, the system appears to be operating filled at all times. The ASE noise source is typically just a regular EDFA amplifier module running without an input signal (open loop), which creates ASE optical noise at all wavelengths across the band.
(Figure 1) ASE Noise Loading on unused channels
Incorporating ASE noise loading simplifies optical power management algorithms since these control algorithms no longer need to be “dynamic,” adjusting for changes in the network as channels are added or deleted. The power management algorithms can be static, only needing adjustment during initial provisioning. Since the system appears filled at all times, no additional optical power adjustments are required – at least in theory. Even with ASE noise loading, there are some failure scenarios where “dynamic power management” may be needed to ensure error-free operation.
On subsea systems, the ASE noise loading is part of the submarine line terminating equipment (SLTE) – the ROADM node installed in the cable landing station. With ASE noise injected into unused channels, the underwater repeaters can incorporate simpler amplifier designs with fewer components, lowering the risks of high repair costs and long repair times.
Noise Loading – Terrestrial Networks
Dynamic power management algorithms have been the primary techniques for over 20 years on terrestrial WDM networks. While there has been some recent industry interest in utilizing ASE noise loading on terrestrial networks, particularly on C+L WDM networks, there aren’t significant performance differences or advantages to either method.
Fortunately, industry WDM vendors support dynamic power management algorithms and integrated ASE noise loading on their iROADM platforms, enabling carriers to choose either approach.
Nonlinear impacts
Optical nonlinearities such as cross-phase modulation (XPM), four-wave mixing (FWM) and self-phase modulation (SPM) are unwanted wavelength interferences and distortions. Nonlinearities result in a slight reduction in overall OSNR performance, slightly reducing a network’s capacity and optical reach. The amount of the nonlinearity penalty depends on several factors, including fiber type, number of active channels, channel spacing, route distance, and wavelength power levels. To ensure networks operate with their designed wavelength capacity and optical reach, vendor WDM simulation tools calculate the nonlinearity penalty and include it in their overall OSNR budget for a network design.
There has been some industry discussion on whether ASE noise-loaded systems provide a more accurate estimate of wavelength OSNR budgets than relying on vendor WDM simulation tools. Since ASE noise loading fills all unused channels, the theory is that noise-loaded systems provide static, known operating performance, including all nonlinear penalties, that doesn’t vary as channels are added or deleted. Historically, on terrestrial WDM networks, dynamic power management algorithms provide per-channel optical power and tilt management. At the same time, vendor WDM simulation tools calculate the OSNR budget and nonlinear penalties to ensure network performance remains constant regardless of the number of wavelengths operating on the network. Both approaches work equally well and result in approximately the same OSNR budgets and overall performance. In anything, there is some industry research that indicates ASE noise loading may be a bit overly conservative as an estimation of OSNR performance.1
C+L Networks
Optical power management on C+L WDM networks is more complex due to stimulated Raman scattering (SRS) affecting both C and L-bands. SRS is a nonlinear effect that causes optical power to shift from shorter wavelengths to longer wavelengths, resulting in wavelength tilt. This effect is present in C-band-only WDM systems, as well as C+L WDM systems, as shown in Figure 2.
With C-band-only networks, power management algorithms automatically correct for SRS-induced tilt; they maintain a flat spectral response across the band. Since the amount of SRS tilt is dependent on the number of wavelengths on the network, the tilt compensation is automatically adjusted as wavelengths are added or removed from the network.
(Figure 2) SRS-induced wavelength tilt
On C+L systems, these dynamic power management algorithms operate across both C and L bands, adjusting the wavelength powers and SRS tilt compensation in both C-band and L-band amplifiers to ensure a flat wavelength spectrum across both bands, as shown in Figure 3.
(Figure 3) C+L dynamic power management
SRS tilt compensation is still required on noise-loaded systems, but the amount of tilt compensation remains constant. The SRS tilt compensation doesn’t vary since ASE noise loading fills all unused channels, simulating a fully loaded network.
One issue of ASE noise loading on C+L WDM systems, especially as part of SRS tilt compensation, is that it typically requires the deployment of both C-band ASE noise sources and L-band ASE noise sources as part of the initial deployment – even if a carrier initially only uses the C-band capacity. As a result, the initial network costs can be slightly higher when using ASE noise loading. In addition, failure of an ASE noise loading source could cause bit errors or traffic outages unless the WDM network also supports C+L dynamic power management as a backup solution.
Industry Myths and Misinformation
There is some industry misinformation that suggests ASE noise loading improves optical restoration times in WDM networks. Noise loading should not impact restoration times – on well-designed WDM nodes.
After a network outage, ROADM nodes carefully control and manage the restoration of dropped channels to prevent power transients, which could cause bit errors on unaffected traffic-carrying wavelengths. In well-designed WDM equipment, typical restoration times range from 5 to 15 seconds, depending on the network size and several channels being restored. ROADM nodes incorporating well-designed optical restoration algorithms incur no impact on restoration times, regardless of whether the network utilizes noise loading.
A network experiencing exceptionally long restoration times may indicate an underlying deficiency in the implementation and design of a specific vendor’s optical restoration algorithms. While ASE noise loading may improve (or mask) a vendor-specific product issue, there’s no generic industry benefit regarding restoration speed.
WDM Power Management
WDM nodes actively manage their systems to ensure flat gain response across all wavelengths with constant per-channel power levels, also known as continuous power spectral density (PSD). Traditionally, terrestrial networks have utilized dynamic power management algorithms running within each ROADM node to automatically adjust optical power levels and ensure optimal performance - at all times.
On C+L networks, the dynamic power management algorithms operate seamlessly across both bands. As an alternative, there is some industry interest in using subsea ASE noise-loading techniques on terrestrial networks, particularly on C+L WDM systems. Power and SRS tilt management are still required on ASE noise-loaded systems, but those adjustments can be “static.” There are no significant performance differences between dynamic power management algorithms and noise loading.
Fortunately, industry vendors support both dynamic power management and ASE noise loading options on their WDM systems, so carriers can choose their network deployment preferences.
[1] S. Searcy, T. Richter, S. Tibuleac, “Experimental Study of Bandwidth Loading with Modulated Signals vs. ASE Noise in 400ZR Single-Span Transmission”, OFC Conference, W1G.6, March 2022
Randy Eisenach is part of the WDM and High-Speed Optics team at Nokia. He specializes in optical transport technologies, next-generation ROADM architectures, and high-speed photonics.