PICs To Play Essential Role In Disaggregating Network Infrastructure
Today’s metropolitan telecom networks employ photonic and microelectronic technologies to move data. Key infrastructure is manufactured by a small group of international companies that ensure interoperability through control of aggregated, system-level BOMs. Disaggregating infrastructure is an emerging ‘ala carte’ concept designed to enable a wider variety of services while lowering costs. Some believe it may also encourage network expansion into underserved markets. According to the OPTICA industry association and researchers from the EU’s PASSION photonics project, future disaggregated networks will need PICs to realize their fullest potential.
MOVING ANALOG and digital data across the globe is complex and expensive. Today's network infrastructure evolved over decades, growing in speed, bandwidth, complexity and power consumption right along with the ever-expanding demand for faster, high-volume data throughput. But is today's infrastructure procurement methodology and network architecture the best ways to move data and serve customers? Some believe the answer is ‘no.'
Achieving maximum uptime and quality of service (QoS) have become international broadband service benchmarks. One way to ensure this is the aggregated approach now used in sourcing components. Under aggregation, major infrastructure providers such as Ericsson, Huawei and Samsung essentially agree to ensure high reliability through a collectivized BOM, meaning in effect that achieving the highest QoS depends on essential network components being manufactured and sourced through the same company.
While aggregation helps ensure reliability, the process also locks an operator into a vendor relationship that is challenging to undue as was seen pointedly when concerns arose in recent years over whether Huawei was placing undetectable ‘back doors' in new 5G network management software that might allow third party eavesdropping or even the ability by an outsider to take over network control. The company continues to deny such claims, yet concerns have led some regulators to require operators in their countries to choose different providers. Some argue disaggregation could have another less obvious benefit: it complicates efforts by bad actors to gain unauthorized network access.
Despite points against aggregation, the system has worked; infrastructure manufacturers are for the most part sensitive to customer needs and requirements. Nevertheless, the current system has also made upgrading and expanding a metropolitan area network (MAN) expensive and time consuming, especially if an operator wishes to offer a service not supported by their equipment vendor's system management software.
Expanding service into rural markets with lower population densities is also an issue since the countryside lacks the same profitability prospects as do high-density urban centers; if a means could be found to lower these costs it could encourage greater access. Today, operators in most countries typically offer rural customers fewer or more expensive services; most operators depend on governmental subsidies to expand into the farthest, most sparsely populated corners of their markets.
The Optica trade association (formerly the Optical Society of America, sponsor of the OFC conference event,) is actively addressing the prospects and challenges of network disaggregation. Optica recently worked with European Union researchers including those taking part in the EU's Horizon 2020 (EU-H2020) PASSION project. PASSION is designed to support the development of future MANs with high transmission capacity, low cost and reduced energy consumption. The project has focused on the network and its key elements, such as transceivers and network nodes, which were developed by different partners.
Josep Fabrega, of the Centre Tecnològic de Telecomunicacions de Catalunya (CTTC) in Spain, was a part of PASSION project team to develop a MAN disaggregation approach. Fabrega is scheduled to present the group's findings during the OFC Conference and Exhibition on 6th March. His presentation will in part discuss proposed network design architectures and ideas for making disaggregation achievable in future network development. PIC Magazine spoke with Fabrega prior to the event to discuss the roles that photonic integrated circuits can play in future network designs. As part of his group's work, the team demonstrated the feasibility of a disaggregated MAN by considering each relevant component. They used sliceable bandwidth/bitrate variable transceivers, which are multi-flow programmable and enable point-to-multipoint connectivity.
“The challenge (with the transceivers) is to have a cost-effective solution,” said Fabrega. “So, the PASSION project developed a photonic integrated solution that involves the use of specifically engineered photonic integrated circuits (PIC) and vertical-cavity surface-emitting lasers (VCSELs) for the transmitter part. The receiver part is relying on, again, specifically engineered PICs for coherent detection.”
Although Fabrega's group does not speak for system operators, the group spoke with many telecom/datacom operators to ascertain the benefits and negatives of aggregation.
“Until a few years ago, equipment vendors and manufacturers were selling complete (closed) network solutions. This means that telcos were publishing RFQs and vendors were proposing solutions that would include equipment and the management software needed for guaranteeing the end-to-end connectivity in a given network. So, all the network elements were aggregated in the sense that telcos had to use the transceivers, nodes, and amplifiers (among other elements) from the same vendor. Of course, this paradigm has its benefits and drawbacks. The main benefit is that the vendor is guaranteeing the performance of the network,” he remarked.
“Nevertheless, the (aggregated) network is less flexible in terms of resource assignment. This is quite critical when telcos are trying to slice their networks in order to provide differentiated services in a single infrastructure. For example, let's assume that they provide two different services that need to match different performance figures. This would need a highly flexible network solution that could be able to assign the right resources to each service. Therefore, telcos are developing and deploying their own management software and would like to have open and standardized interfaces for the different network elements. That is why the disaggregated approach is an interesting paradigm that telcos are exploring,” he said. Being free to source network equipment from a larger variety of providers enables an operator, for example, to offer types of services that most networks cannot accommodate today. It is about flexibility and innovation, he said.
“A main innovative network function could be network slicing. Let's assume that there are two services: One, critical infrastructure monitoring (e.g. in a railway network) and two, media content delivery (e.g. Amazon Prime Video, Netflix, or similar). These two services have different needs (and priorities) that should be matched. For example, the first service would be asking for limited capacity channels, but would need a consistent/limited latency/delay in order to communicate a plurality of sensors with an IT/cloud datacenter to process all the data and communicate with a command center that would be monitoring the infrastructure.”
“On the other hand, our second example service would ask mainly for high-capacity connectivity between a content delivery datacenter and the corresponding subscribers. Of course, this service would also desire low latency, but this would be at a lower priority than what the first service needs. In order to match the different needs of those services, a telco can setup two different network slices, ensuring an exclusive use of resources (e.g. channels, network paths, transceivers, etc.,) is assigned to each service. In order to be able to perform this assignment at low level, the equipment should be able to be highly flexible and configurable. Of course, this could be achieved either by asking for an increased flexibility to the equipment vendor (increasing the cost of the solution) or by approaching the disaggregated paradigm,” he explained.
Fabrega was quick to point out that the PASSION consortium of research groups dealt with the development of technologies, devices and systems that can support a disaggregated network. But network operators participated, and Fabrega pointed to the fact that Telefonica was a co-author of his group's report, ‘In this sense, Telefonica was proposing the directions to follow and the requirements to be met in order to approach the disaggregation paradigm.” According to Fabrega, designing a disaggregated network will need standardized components that meet internationally developed performance specifications as do components powering today's networks.
“Disaggregation is a paradigm that can be fully achievable with common state-of-the-art pluggable components, provided that they have an open interface and enough flexibility. In our case, we only focused on demonstrating the potential that PICs might offer, since they could cope with network disaggregation requirements while minimizing size, weight and power consumption. Even this last part is not a strong requirement for achieving network disaggregation, but it is of paramount importance in supporting environmental goals,” he said.
Fabrega noted that PICs provide a compact/integrated version of what are today typically bulky devices. Saving space and reducing power consumption are essential PIC qualities. Concurrently, VCSELs offer the advantages of low cost and optical efficiency within a small footprint. VCSELs are lower cost due to a more straightforward manufacturing process than traditional commonly utilized semiconductor lasers.
Since integrated photonic network components are still in their relative infancy compared to existing pluggable network counterparts, Fabrega said that anyone reading his group's report should appreciate that the idea was to demonstrate feasibility, which he believes was accomplished. But at no time did the group attempt to offer up a disaggregated network with fully qualified components ready for city-scale deployment.
“First, new components were developed/manufactured to this purpose: high-speed C-band VCSELs; a polymer optical switch, and an integrated coherent receiver (along with) others that were not used in the experiments. Also, new systems were built based on these devices: network nodes and transceivers. Second, we had to integrate everything in a single place. In this case, it was at the CTTC premises in the Barcelona area of Spain. Since components were manufactured and tested in different areas of Europe (Germany, Netherlands, Italy, and Finland to name a few, and in South Korea,) they had to be appropriately packaged and shipped to the integration facilities,” he explained. The actual integration phase started in mid-2020, which overlapped with the COVID-19 pandemic, a fact that slowed progress at a number of points and delayed completion by a number of months. Fabrega explained that for the network nodes, the researchers proposed an architecture depending on hierarchy level (HL). The HLs can be visualized as concentric rings, where traffic is aggregated from lower to upper levels. For example, traffic coming from HL4 is aggregated at HL3. In turn, HL3 traffic is aggregated at HL2, and so forth. Each transition means that more traffic must be supported.
“For HL4 nodes, PASSION proposes an architecture based on arrayed waveguide gratings in combination with semiconductor optical amplifiers,” said Fabrega. “For higher hierarchy levels, a solution is proposed combining different types of wavelength selective switches together with innovative polymer switching stages.”
The team tested their MAN with a transmission experiment and demonstrated a capacity of 1.6 terabytes per second, which is modest compared to the possible performance of the network, but demonstrates the feasibility of the concept. In the future, the researchers believe some of the constituent devices could be further improved to provide more flexibility. Although Fabrega said he believed the group demonstrated disaggregated network feasibility and its flexibility to enable new provisioning capabilities, he is not certain that disaggregation will automatically lead operators to see advantages to expanding their networks into underserved communities. But he noted that the disaggregation process, if taken to full implementation, could reduce equipment costs to the point that operators may have additional capital resources that could be used to serve new areas. At the same time, he expects that expanding service to some regions will likely still require governmental incentives.
“To my knowledge, it (would be) very difficult that disaggregation could enable (rural) expansion. The aim of disaggregation is to avoid vendor lock-in, and thus increase competitiveness and lower the cost of equipment. On the other hand, the main problem of rural areas is deploying optical fiber cables, rather than equipment. In those cases, a population is usually spread over a wide area. So, in order to cover rural areas, a high amount of work (and cost) is needed to lay down the cables; costs are high and potential revenues are low compared to an urban area where a high number of network subscribers can be expected in a smaller geographic area. This said, network disaggregation can decrease the cost of equipment. This could be beneficial in some specific cases, maybe some villages, in the sense that there might be a business case from the telco point of view. In other words, the decrease of cost in equipment could help to balance TCO in some cases,” he concluded.
Whether disaggregation becomes a driving force in the development of future broadband networks remains to be seen. The concept does offer network operators the chance to remain ‘unbound' to a single infrastructure equipment vendor, which could lead to reduced costs. By way of engaging more and different vendors in the equipment purchasing process, disaggregation could also support more flexible service provisioning. But opening up network equipment sourcing will not by itself provide more cost-effective means to serve rural areas. The build-out of rural broadband networks is likely to continue to require governmental supplements even if disaggregation moves to the center of creating new network architecture in the years to come.
Figure 1: The experimental setup, including all components except the HL3 nodes, demonstrates the feasibility of the disaggregated metro area network.