Hybrid Wins Out in Switching

But while some in the design community feel they are hugely different, some members of the community are starting to view these switching architectures as complimentary. Specifically, designers can combine these switching solutions to develop agile optical networking architectures that provide photonic bypass for connections not requiring electrical processing as well as automated end-to-end connection setup and tear down through some form of electrical switching architecture. Let's look at these architectures in more detail below.

Architecture options
There are three architectures that can be employed when developing agile all-optical networking systems. The first is an agile electrical overlay architecture that provides agility via electrical switches only, while photonic bypass is used for cost reduction. In this case, the photonic layer is static (or manually configurable) [Figure 1a].

The second is an agile photonic and electrical network. In this architecture agility is provided at both the electrical and photonic layers (Figure 1b).

The final is an agile photonic-only network. This architecture only includes photonic agility, leaving the electrical switching section static (Figure 1c).

Clearly, the challenge for a designers looking to develop hybrid photonic/electrical switches lies in choosing the right architecture. Let's look at some challenges designers will face when making this decision.

Today, there are a lot of engineers looking at the electrical-only approach highlighted in Figure 1a above when building agile optical networking architectures. But, the electrical-only approach provides some distinct challenges for designers.

The first challenge for the electrical-only approach is that it does not support selective regeneration, or the capability to use a small pool of regenerated optical-electrical-optical conversions (OEOs) for a larger set of wavelength resources. Thus, a wavelength is regenerated only if this is needed, depending on the route that the connection takes.

Another challenge facing electrical-only systems is the fact that they do not allow for redirection of OEO resources from one direction to the other and thus does not adequately support changes in the traffic pattern from the originally projected traffic. Electrical-only architectures also don't support low-cost restoration of wavelength services, the dynamic connection of a wavelength to a test set for trouble shooting purposes, access to all the bandwidth on the line without having to populate OEOs for all wavelengths, and support for automated re-optimization of the network.

The photonic-only method provides its own set of headaches for designers. First, the photonic-only method does not deliver support for aggregation of low-end connections that cannot be cost-effectively carried over an entire wavelength. It also does not offer support for hitless bridge and roll of services from one path to the other. This functionality requires dynamic bridging of the signal and quick switchover to the new path that only electrical switching can provide to date. Additionally, photonic-only architectures fail to provide support for Sonet-like fast protection switching and flexibility in how OEOs are connected to clients.

The Hybrid Advantage
The issues faced by the 'only' architectures described above clearly make the combined photonic/electrical switching architecture a more attractive option for designers. There are several reasons why.

Naturally, having both switching technologies (architecture b) allows the network to enjoy the benefits of both other architectures. For example, the combined architecture can deliver selective regeneration while supporting the aggregation of low-end connections.

What's even more interesting are the capabilities the hybrid approach delivers beyond the ones provided by the 'only' solutions discussed above. For example, the hybrid architecture allows designers to flexibly connect OEOs on both the client-facing side and line-facing side. Thus the OEOs can be referred to as a pool of "floating" shared resources that can be used for any client as well as for any wavelength. This allows for the following features:

1. An OEO can be used for two different purposes, normally requiring two separate devices. It can be used as an adaptation device, taking the client signal and converting it to the appropriate WDM signal (a.k.a. on/off ramp OEO). At the same time, it can be used together with another OEO as a regenerator, by cross-connecting two OEOs by the electrical cross-connect. The merging of such two pools of resources into one pool allows for better efficiency and reduces the dependency on accurate forecasts when designing the network.
2. The photonic/electrical architecture allows designers to combine simple electronic protection schemes, such as Sonet rings, with the flexibility of photonic mesh networking. Therefore, designers can build equipment that supports "virtual rings", or rings whose nodes and the links between them can be configured remotely to better fit the traffic.
3. A related feature, that serves to enhance the protection scheme at the electrical layer, is photonic restoration. This function kicks in after a failure has occurred as a second tier mechanism to enhance the protection scheme and prepare the network for another failure.
4. Efficient and simple support for 1:N protection against failures of OEOs requires agility on both client and line sides, to allow a client signal to be redirected to a different OEO that would feed into the same wavelength as the failed OEO.

Now that we've described why it's better, let's take a deeper look at how the hybrid architecture works.

The Hybrid Node
The hybrid photonic/electrical switching architecture is developed using a nodal architecture incorporating both electrical switching (EXC) and photonic switching (PXC) as shown in Figure 2. Note that this functional description does not preclude the integration of OEOs into the electrical crossconnect function as a further cost reduction measure. Also note that the PXC function can be performed by a large NxN photonic switching matrix or by more limited combinations of wavelength plane switches or wavelength selective switches, as long as they provide enough flexibility to avoid blocking.

As Figure 2 points out, the nodal architecture allows for a small pool of OEOs to be flexibly used to serve a larger number of potential clients and an even larger number of potential wavelength resources. Photonic passthrough is achieved by switching the signal at the PXC layer, whereas regeneration is achieved by switching the desired wavelength to an OEO at the PXC layer and connecting it to another OEO through the EXC.

Pre-Deployment of OEO Resources
Network agility implies that the relevant resources to support the next connection request must be in place beforehand and thus the cost of the network must always be higher than the absolute minimum needed for the current level of traffic. This phenomenon is called "pre-deployment" of resources (or "over-provisioning").

Since photonic layer resources—in particular OEOs—are expensive, network agility has a capital expense (capex) implication that to some degree offsets the operational expenditure (opex) advantages that agility promises. Thus, minimizing the per-deployment is key to agile networking.

This problem is not hard to solve given accurate forecasts, since the pre-deployed resources are guaranteed to be eventually used optimally, when the traffic grows as planned. The problem, however, is that accurate forecasts do not exist, especially with the changes in communication usage patterns that have occurred in recent years.

The challenge designers face is minimizing pre-deployment costs in face of inaccurate forecasts. This is hard to do without photonic agility, because the desire to have points of agility everywhere for an efficient agile network conflicts with the desire to have as much photonic bypass for the lowest cost solution. As a result, the more use is made of photonic bypass, the more lightpaths are required to connect different nodes, which translates into more OEOs to terminate those lightpaths. We refer to this as the "pre-deployment explosion" phenomenon, which is a dominant component of the cost of this network architecture (see Figure 3).

As shown in Figure 3, inaccurate traffic forecasts are handled better if the photonic layer is agile, as opposed to electrical agility. The pool of OEO resources becomes an aggregated nodal pool, as opposed to a separate pool for every photonic adjacency of the node.

The move from per-adjacency forecasts to nodal forecasts reduces the dependency on their accuracy and reduces the number of pre-deployed resources assuming imperfect forecasts. Even more importantly, it simplifies the planning process for the carrier, which in turn has a potential to further reduce opex.

Dealing with the Cost Issue The discussion above focused on the benefits of the combined photonic/electrical switching approach. However, cost is one issue that designers must grapple with when implementing this hybrid architecture.

Critics will say that combining both electrical and photonic switching elements in the same architecture comes at a cost premium. In reality, however, if all network factors are taken into account the cost reduction associated with adding the photonic switching elements to existing electrical elements could offset the additional costs designers might incur when implementing the hybrid architecture.

Editor's Note: This paper is based on a presentation made at the 2002 Communications Design Conference.

About the Author
Ori Gerstel is an independent network architecture consultant based in the Silicon Valley, CA. Until recently he was a senior systems/network architect at Nortel Networks. Ori holds a Ph.D. degree from the Technion, Israel. He can be reached at ori@ieee.org.