by Richard Jensen and Andrew Lord
The recent explosion in access bandwidth for Internet and other services is driving carriers to provide a greater level of optical layer flexibility in regional and long-haul networks. The flexibility is needed to solve a wide variety of issues plaguing carriers and service providers. They urgently need to further automate and accelerate provisioning of customer services along with providing new types of revenue-generating wavelength services. In addition, they need to lower both capital and operating costs while improving network resiliency and availability.
The solution is to capitalise on the power of optics to inexpensively switch large amounts of bandwidth at the physical layer, and operators are taking first steps towards this by installing reconfigurable optical technologies. Wavelength-selective switch (WSS) technology is a key building block that enables carriers to meet these diverse needs. The technology has matured to the point where WSSs are poised for large-scale deployment in long-haul and even regional networks.
A WSS is a component in a network switching node that is able to switch individual DWDM traffic wavelengths dynamically among different fibres without any optical-to-electrical (OE) conversions. Todayâ��s WSSs are typically 1Ã�5 or 1Ã�9, which means they have one input fibre and five or nine output fibres, respectively. Since WSSs are all-optical devices, they can also be used in reverse to switch waves from multiple input fibres to a single output fibre. To build up a network wavelength switch, WSSs can be interconnected together to make a nonblocking wavelength switch among many fibres.
Recent studies have shown that the WSS specifications, especially loss, are critical to the network economics and performance. Emerging WSS technologies promise to lower this loss from the current 6- to 9-dB range to less than 3 dB. For example, a WSS based on piezoelectric technology that is currently in final development is projected to have less than 3 dB of optical loss.
A comprehensive study of a transparent WSS backbone using a London regional reference network showed that even small reductions in WSS loss can dramatically lower overall network losses, thereby improving performance and reducing costs. The concept of â��cost of lossâ�� was used to quantify how WSS performance affects cost. The results of this London regional network study are applicable to a wider range of regional, long-haul, and larger metro networks and show the dramatic impact lower WSS losses can have on network performance and cost.
To quantify the impact of lower WSS loss, we studied adding a reconfigurable optical layer using WSSs to a possible London regional network. This real-world network has many attributes, such as a large number of nodes and diverse traffic patterns, that make it an ideal case study for understanding the impact of WSS loss on a wide range of regional and long-haul network applications. The reference model has 38 switching nodes with connectivity degree 2 to 6 and 62 interconnecting transport links, and a realistic traffic model was assumed.
The overall network is shown in Fig. 1. For each transport link, we took into account the number of wavelengths, wavelength assignments, fibre lengths, loss, chromatic dispersion, and polarisation-mode dispersion (PMD). For each switching node, we took into account the number of interconnection fibres and the amount of add/drop traffic terminating at the node.
We introduce the concept of â��cost of lossâ�� to understand the impact of higher WSS loss on the overall network. First we calculate the total loss impact that the WSS introduces in the network on both the add/drop and through traffic, and then we determine a cost penalty based on the amplification costs required to overcome this loss. There are several ways to translate this loss into network costs. For this study we used typical costs seen by service providers for optical amplification, which is about $1,670 per decibel of gain.
In this study we compare the impact of through traffic and add/drop WSS losses of 3, 6, and 9 dB on the overall network. Legacy WSS and waveblocker technologies can have losses up to 9 dB or more depending on number of fibres that are intended to interconnect in the node. WSS technologies available today typically have losses in the 6- to 9-dB range while emerging WSS technologies are expected to be less than 3 dB. For loss values and costs of other network components used in this study, we used typical industry norms.
The study shows that lowering the WSS loss can yield significant network savings and performance improvements. Each additional 1 dB of WSS loss adds about 450 dB to the total loss of the London regional network. The loss due to the WSS was calculated for each node based on the number of through paths and the number of add/drop traffic channels. The overall network loss is calculated by adding up the individual node losses.
Fig. 2 shows a comparison of the overall contribution of the WSS to the total network for WSS losses of 3, 6, and 9 dB. This clearly demonstrates that even small amounts of extra WSS loss can add thousands of decibels of total loss to the overall network. The 6- and 9-dB loss WSSs add an extra 1,332 dB and 2,664 dB of loss, respectively, to the overall network. These figures are so high because the WSS is located in the core of the network, and every signal passes through a WSS when entering or exiting the network and passes through two WSSs in every node while traversing the network. In this particular network, the average node degree is 3.3, giving on average 6.6 WSSs per node and about 250 WSSs in total.
There are several ways to compensate for the extra WSS loss in the network. Regardless of how it is done, making up for these large amounts of extra WSS loss is expensive. Simply increasing the optical amplifier gain and output power will compensate for the loss but will increase the cost of the amplifiers, increase the nonlinear impairments, and reduce overall system reach and performance. Alternatively, adding extra amplifiers would increase amplification. This alternative adds considerable cost since the amplifiers need to be powered, monitored, and maintained. Also, adding a large number of extra active components like optical amplifiers lowers the overall network reliability and availability. Simply increasing the transmitter output power or receiver sensitivity would also add cost and could have performance implications.
These extra WSS losses add significantly to overall network costs. Using a typical service provider optical amplifier cost of $1,670 per decibel of gain, loss compensation adds about $750,000 per decibel of WSS loss. For the 6- and 9-dB WSSs cases, compensation adds over $2.25 million and $4.5 million, respectively, above the cost of using the 3-dB WSS. In the case of the 9-dB WSS, the cost of compensating for the WSS loss is higher than the overall WSS equipment costs.
Unlike optical fibre losses and other component losses that cannot be lowered due to material properties, physical laws, or cost constraints, there is no fundamental reason why WSS optical losses cannot move lower in the near future. In addition to saving money in near-term deployments, the emerging ultralow-loss WSS technology also enables innovative architectures that are not currently possible with higher-loss technology.
The impact of WSS loss in this study of the London regional network showed significant savings that can be achieved by even small reductions in WSS loss. The metric of â��cost of lossâ�� quantifies the impact of WSS loss on network performance and cost. This study shows that for higher-loss WSS technologies the cost of using these WSSs in the network can be comparable to the overall WSS equipment costs.
The results of this London regional network study can be applied to a wide range of regional and some long-haul network deployments. Also, the underlying traffic patterns and connectivity of this real-life network make a powerful statement about the impact of WSS technology in typical network deployments. Using lower-loss WSSs directly translates to lower bottom line network costs. As WSS and other component costs continue to come down over time, the value of lower WSS loss will become increasingly important to the viability of network deployments.
Richard Jensen is director of network architecture at Polatis (www.polatis.com). He can be reached at firstname.lastname@example.org. Andrew Lord is responsible for optical core R&D at BT (www.bt.com). He can be reached at email@example.com.