The IEEE website places cookies on your device in order to provide you with the best user experience. By using our website, you agree to the placement of these cookies. To learn more, please read our privacy policy.
For local area networks or campus data networks, installing optical fibers with thick cores once seemed a good idea. It is much easier to couple light into this "multimode" fiber than into the thin core of a high-capacity "single-mode" fiber used in long-distance networks.
The thicker the fiber core, the slower the data flow through the fiber, but a fiber with a 50 micrometer (µm) core can transmit data at a rate of 100 megabits per second, and the transmission distance can reach hundreds of meters-enough for local transmission.
Now Cailabs, located in Rennes, France, has developed special optical devices, which are said to be able to send signals at 10 gigabits per second (Gbps) through the same fiber, up to 10 kilometers long, without the need to replace traditional multi-mode optical fiber. They hope to reach a rate of 100 Gbps, which is now widely needed in large data centers.
Determining the speed of data transmission through an optical fiber depends on the core size of the optical fiber because the core acts as a waveguide. The long-distance fiber has a 9-micron core, which limits the 1.55-micron infrared light used in it to a single narrow transmission mode.
Multimode cables with a core diameter of 50 or 62.5 µm allow light to propagate in many different modes. However, these modes follow different paths, so the time required to travel through the fiber is slightly different. Therefore, as light propagates through the fiber, the timing of many pulses spreads-an effect called modal dispersion, which limits the bandwidth of multimode fiber. In the early days of optical fiber, engineers accepted the limited bandwidth of multimode fibers as a trade-off because their larger cores greatly reduced the optical loss at many connections in the local network.
Single-mode fiber began to be used for long-distance transmission in the 1980s. Operators need to achieve data rates of hundreds of megabits per second (Mbps) within tens of kilometers, and improvements in splicing and connector accuracy have made single-mode fiber practical over long distances. Today, single-mode fiber can transmit 100 Gbps over up to 100 closely spaced wavelengths, for a total of 10 terabits per second.
Multimode fiber is still the preferred option for local and campus networks because it is cheaper and easier to install where many connectors and reconfiguration are required. For Ethernet that transmits 100 Mbps, modal dispersion can still be tolerated within a few kilometers.
However, Ethernet transmission at gigabit per second speed requires special adjustments to multimode fiber to reach 550 meters, while Ethernet that provides 10 Gbps is limited to shorter distances. This means that operators of local area networks and campus networks must upgrade their installed networks to GigE or 10-GigE.
In many cases, the obvious choice is to replace the old multimode fiber network with single-mode fiber. If the old cable is easily accessible, this is relatively easy and cheap. However, anyone who has ever had to rewire an old house knows that if rewiring requires heavy construction or passing new cables through twisted paths in existing walls, it can cause damage and be very costly. Founded in 2013, Cailabs is applying the technology it has developed to shape the beam very precisely. This technology aims the beam at the optical fiber so that essentially all of the light is captured in one of the multiple modes that the optical fiber can carry.
"Basically, it's all about introducing the right model to fiber optics," said Jean-François Morizur, the company's chief executive officer.
Since all light is in one mode, the pulse timing of modal dispersion is not spread, and multi-mode fiber can transmit at a much higher data rate than single-mode fiber. "You have to be very precise, so 99.5% of the light will be in the correct mode, which is difficult to do," Morizur said.
The company did not provide detailed information about its process, but telecommunications researchers seeking to achieve ultra-high fiber capacity have transmitted signals in up to six different modes through fibers with a core of about 20 µm. Each of these modes can be transmitted at a single-mode rate, allowing Modular Division Multiplexing, multiplying the data rate by the number of modes used.
Cailabs has packaged this technology in its Aroona system, which can be plugged into an existing multi-mode network, increasing capacity by a hundredfold. The company has installed many systems in universities, chemical plants, defense bases, oil refineries, and high schools in Europe and the United States.
A few weeks ago, Cailabs upgraded a multi-mode system spanning four kilometers for a large chemical plant in Germany. "Now we have reached the eligibility criteria of 10 Gb [per second] and are testing 100 [Gbps]. You can imagine the cost savings compared to rewiring at the factory," said Morizur.
Georgia Institute of Technology found that the cost of replacing the 400 to 1,100-meter cable with 35 independent sorority houses was unaffordable, but by plugging in the Cailabs module, it was able to upgrade the original 100 Mbps capacity to 10 Gbps over the weekend. Richard Mack, a principal analyst at the market research firm CRU Group, estimates that since 1980, approximately 8.5 to 90 million kilometers of multimode fiber have been installed. Some are not upgradeable because they are built into dedicated links or used in non-networked applications.
Jeff Hecht writes articles about lasers, optics, fiber optics, electronics, and communications. He has received engineering training and is a lifelong senior member of IEEE. He likes to study the working principles of laser, optical and electronic systems, and explain their applications and challenges. Currently, he is exploring the challenges of integrating lidar, cameras and other sensing systems and artificial intelligence in autonomous vehicles. He recorded the history of laser weapons and optical fiber communications, and wrote tutorial books on lasers and optical fibers.
We have all been told that 5G wireless will provide amazing features and services. But it will not be cheap. All in all, in the next five years, the deployment of 5G will cost nearly US$1 trillion. This huge cost will be mainly borne by network operators, AT&T, China Mobile, Deutsche Telekom, Vodafone and other companies, as well as dozens of companies that provide cellular services to customers around the world. Faced with such a huge cost, these operators have asked a very reasonable question: How can we make it cheaper and more flexible?
Their answer is: You can mix and match network components from different companies to promote more competition and lower prices. At the same time, they have caused a split in the industry on how to build wireless networks. Their opponents—sometimes even reluctant partners—are the few telecom equipment vendors who can provide the hardware that network operators have been buying and deploying for years.
These vendors initially opposed the plan called Open RAN because they believed that if implemented, it would damage—if not disrupt—their existing business model. But in the face of the collective power of operators who demand new ways to build wireless networks, these providers have few options, and none of them are very attractive. Some people responded by trying to set conditions for the development of Open RAN, while others continued to delay and risk being left behind.
It may take a decade or more for technologies that support a generation of wireless technologies such as 5G to go from the initial idea to the full realization of the hardware. In contrast, Open RAN appeared almost overnight. In less than three years, this idea has gone from a concept to multiple major deployments around the world. Its supporters believe that it will nurture huge innovations and reduce the cost of wireless access. Its critics say it threatens basic network security and may cause disaster. Either way, this is a watershed in the communications industry, and there is no turning back.
Rakuten Mobile's Open RAN network includes Nokia 4G radios running software from other vendors. The company has deployed one such RAN at its global headquarters in Tokyo. The Open RAN network also uses servers to power the cloud-native network. Photo: Lotte
Broadly speaking, a radio access network (RAN) is a framework that connects terminal devices such as mobile phones with a larger wired core network. Cellular base stations or towers are the most common example of RAN. Other types of base stations, such as small base stations that send and receive signals over short distances in 5G networks, also meet the requirements.
To act as this link, the RAN performs several steps. For example, when you use your mobile phone to call a friend or family member in a different city, you need to be within the range of the mobile phone tower. So the first step is to let the antenna of the cell phone tower receive the cell phone signal. Second, the radio converts the signal from an analog signal to a digital signal. Third, a component called the baseband unit processes the signal, corrects the error, and finally transmits it to the core network. In the RAN, these components—antennas, radios, and baseband units—can and are often regarded as discrete technical blocks.
If you separate the radio and baseband unit from each other and develop and build them independently, you still need to make sure they work together. In other words, you need their interfaces to be compatible. Without this compatibility, when moving from radio equipment to baseband equipment, data may appear garbled or lost, and vice versa. In the worst case, radio and baseband units with incompatible interfaces will not work together at all. Functional RAN requires a common interface between these two components. Surprisingly, however, there is currently no guarantee that a radio made by one supplier can interoperate with a baseband unit made by another supplier.
Like all standards for cellular networks, the specifications for RAN interface standards are formulated by the third-generation partner program. Gino Masini, chairman of the 3GPP RAN3 working group, said that many 3GPP specifications, including those covering interfaces, are designed with interoperability in mind. However, Masini, who is also the lead researcher on standardization at Ericsson, added that nothing prevents vendors from "complementing" standardized interfaces with additional proprietary technology. Many vendors do this - and Masini says this does not limit vendor interoperability.
Others in the industry disagree. "Nokia and Ericsson are using 3GPP interfaces that should be standard," said Eugina Jordan, vice president of marketing at Parallel Wireless, a company that develops open RAN technology in New Hampshire. But "these interfaces are not open, because each supplier will create its own flavor," she added. Most of these vendor-specific adjustments occur in the software and programming language used to connect the radio to the baseband unit. Jordan said that these adjustments are mainly in the form of supplier-defined radio parameters, which are intentionally left blank in the 3GPP standard for future development.
Ultimately, this will cause the hardware built by each vendor to be incompatible with hardware from other vendors, and it will not make operators comfortable. "We are seeing a widening gap in the 3GPP specifications," said Olivier Simon, director of radio innovation for French operator Orange. Open, because they cannot achieve multi-vendor cooperation on both sides of the interface. "
The O-RAN Alliance of which Simon is a member of the Executive Committee is the largest industry organization engaged in the Open RAN specification. The group was established in 2018 when five operators including AT&T, China Mobile, Deutsche Telekom, NTT Docomo and Orange joined to lead the development of more industries in Open RAN. Sachin Katti, associate professor at Stanford University and one of the co-chairs of the O-RAN Alliance Technical Steering Committee, said: "I think we need to create a unified global operator voice to promote this decomposition and openness."
The members of the O-RAN Alliance hope that Open RAN can fill the gap created by the 3GPP specifications. They quickly said that they were not going to replace the 3GPP specifications. Instead, they see Open RAN as a necessary tightening of the specification to prevent large vendors from applying their proprietary technology to the interface, thereby locking wireless operators into a single vendor network. By forcing open interfaces, the wireless industry can find a new approach to network design. If those open interfaces can promote more competition and lower prices, so much the better.
With the early deployment of 5G on a global scale, in 2019, the GSM Association, a wireless industry organization, predicts that operators will spend US$1.3 trillion on 5G infrastructure, equipment, and technology for their networks. RAN construction will consume most of these capital expenditures. Most of the expenditure will be spent on a few vendors who can still provide a complete end-to-end network.
"This has always been a pain point because RAN is the most expensive part of an operator's deployment," said Sridhar Rajagopal, vice president of technology and strategy at Mavenir, a Texas-based company that provides end-to-end network software. "It requires nearly 60% to 70% of the deployment cost." The GSM Association predicts that by 2025, operators will spend up to 86% of their capital budget on the RAN.
Cellular networks use wired or fiber optic backbone networks called core networks to send signals over long distances. The radio access network (RAN) acts as an intermediary, receiving wireless signals from mobile phones through antennas, converting the signals into numbers in the wireless unit, and performing tasks such as data processing and data processing, and connecting terminal devices such as mobile phones to the core network. Error correction in the baseband unit. In the current 5G system, the baseband unit splits these tasks between the distributed unit and the centralized unit. The open RAN concept hopes to build on this division to create a more flexible and thinner RAN.
Not surprisingly, with so much money, operators will do everything they can to avoid any fiasco caused by hardware incompatibility. The most reliable way to avoid this disaster is to insist on using the same provider from one end of the network to the other, so as to avoid any possibility of mismatched interfaces.
Another factor that makes operators uneasy is that there are fewer and fewer companies that can provide cutting-edge end-to-end networks. Now there are only three: Ericsson, Nokia and Huawei. These three end-to-end providers can charge high prices because operators are basically locked in their systems.
Even with the arrival of a new generation of wireless technology, it will not create a clear opportunity for operators to switch suppliers. The new generation of wireless products maintain backward compatibility, so, for example, when a 5G mobile phone is not within the coverage of any 5G cell, it can run on a 4G network. Therefore, as operators build their 5G deployments, most of them insist on using a single vendor’s proprietary technology to ensure a smooth transition. The main alternative is to scrap everything and pay more for the new deployment from scratch.
The wireless industry has reached a broad consensus that Open RAN makes it possible to select different RAN components from different vendors. This opportunity, called decomposition, will also remove the pressure on whether components will cooperate when they are inserted. Whether decomposition is a good thing depends on who you ask.
The operator definitely likes it. Dish, a TV and wireless provider, is particularly active in adopting Open RAN. Siddhartha Chenumolu, vice president of technology development at Dish, described his first reaction to the technology: "Hey, there may be something here that we can completely decompose," he said. "I don't have to rely only on Ericsson to provide radios, or only Nokia." Dish promised to use Open RAN to fully deploy 5G networks in the United States this year.
Proponents of Open RAN are exploring several possible "functional splits" to create new, interoperable interfaces in the RAN system, of which four possibilities have received the most attention. Each split allocates many tasks undertaken by the RAN to create a link between the core network and the terminal equipment based on different ways that different types of cellular networks may require. For example, Split 2 creates a highly intelligent radio unit that handles most of the data processing before signal transmission. On the other hand, Splits 7.2x and 8 create "dumb" radios that minimize data processing to reduce latency.
Smaller, more specialized vendors are also optimistic about the boost Open RAN can bring to their businesses. For Software Radio Systems, an advanced software-defined radio manufacturer, Open RAN makes it easier for them to focus on developing new software without worrying about losing potential customers because of integrating the technology into a wider network.
Not surprisingly, the remaining three hardware vendors hold different views. In February of this year, Franck Bouétard, CEO of Ericsson France, called Open RAN an "experimental technology" that is still several years away from maturity and cannot compete with Ericsson's products. (Ericsson declined to comment for this article).
However, some industry insiders believe that hardware manufacturers deliberately slow down the development of Open RAN. Paul Sutton, director of Software Radio Systems, said: "Some large vendors keep asking one question or another. Ericsson may be the most resistant to Open RAN because they may lose the most."
Not every major supplier is fighting back. For example, Nokia sees an opportunity. "I think we need to accept the fact that Open RAN will happen with or without us," said Thomas Barnett, Nokia's head of mobile network strategy and technology. Occupy a leading position to seize a better market share. "For example, Japanese operator Lotte is using Nokia's equipment for its Open RAN deployment. Nokia is also cooperating with Deutsche Telekom to deploy the Open RAN system in Neubrandenburg, Germany, later this year.
This is not to say that Nokia or other vendors are on the same page as operators and professional vendors such as software radio systems. At present, there are still many controversies. Ericsson and other vendors believe that creating more open interfaces will inevitably create more cyber attack points in the network. Operators and other Open RAN proponents countered that standardized interfaces will make it easier for the industry to identify and fix vulnerabilities. Everyone seems to be open enough to be open enough, or have different opinions on how much RAN hardware elements should be broken down.
In its most ambitious version, Open RAN splits the RAN into smaller components besides the radio and baseband unit. Proponents of this level of decomposition believe that by allowing the company to be hyper-specialized, it will bring more suppliers into the wireless industry. For example, an operator can sign a contract with a supplier to only prepare the processor for data received from the core network for wireless transmission. Many industry insiders also said that this specialization can accelerate technological innovation by replacing and deploying new RAN components without waiting for the entire radio or baseband unit to be upgraded. "This may be one of the brightest opportunities Open RAN can provide," said Tedlar Papport, the founding director of the New York University Wireless Technology Research Center.
The wireless industry’s first effort to disaggregate was inspired by the 5G specification itself. These specifications split the baseband unit into two smaller components, which are responsible for processing data transmission and data transmission with the core network. A component is a distributed unit, which assumes the responsibility of data processing. The other component is the central unit, which handles the connection to the core network. The advantage of splitting the baseband unit in this way is that the centralized unit no longer needs to be located in the cell tower itself. Instead, a centralized unit can be located in a local server farm, maintaining connections to the core network for multiple cell towers in the area.
The O-RAN Alliance is studying some different "functional splits" in RAN to create more splitting opportunities beyond this split between distributed units and centralized units. Each of these additional splits creates a partition somewhere in many steps between the arrival of the signal from the core network and the transmission to the phone. It's a bit like a lunch break: you can have lunch early, thereby shifting many of your responsibilities to the afternoon, or work for a few hours and choose a later lunch.
An important split called Split 7.2x delegates the responsibilities of signal encoding and decoding and modulation to the distributed unit. On the other side of the split, the radio is responsible for some light processing tasks, such as beamforming, which determines the specific direction of transmission. The radio is also responsible for converting digital signals into analog signals and vice versa.
Another split, Split 8, even shifted the responsibility of beamforming to the distributed unit, leaving the radio only responsible for converting the signal. In contrast, Split 2 will transfer encoding, decoding, modulation, beamforming, and even more processing responsibilities to the radio, allowing the distributed unit to only compress the data to a smaller number of bits, and then transmit the data to the centralized式units.
The goal of creating open standards for multiple splits is that operators can purchase customized components for the specific types of networks they are building. For example, an operator may choose Split 8 for large-scale deployments that require a large number of radios. This split allows the radio to be as "clumsy" as possible and therefore as cheap as possible, because all processing takes place in the central unit.
Technically speaking, it is possible to combine decomposed RANs with open interfaces using only hardware, but there are some advantages to defining components in software. "Our industry has become very, very hardware-centric," said Chih-Lin I, who served as co-chair of the O-RAN Alliance Technical Steering Committee with Katti of Stanford University. "Every generation of our network basically relies on the close integration of special-purpose hardware and software. Therefore, every time we need an upgrade, a new version, or a new partial version, it takes years."
In order to get rid of the hardware-centric attitude, the O-RAN Alliance also encourages the wireless industry to integrate more software into the RAN. Software-defined networks use programmable software equivalents to replace traditional hardware components and are more flexible. Upgrading virtual components is as simple as pushing new code to the base station.
The emphasis on software also makes it possible for the industry to consider new technologies, the most important of which is the RAN intelligent controller. RIC collects data from the RAN components of dozens or hundreds of base stations at a time, and uses machine learning technology to reconfigure network operations in real time. It is modified based on whether a particular cell tower is under heavy traffic load, for example, or transmission during heavy rain that may weaken the signal. RIC can reprogram the software components of RAN to provide better services. Dish’s Chenumolu said: “Imagine that I can really adjust my network based on user experience and user’s real-time feelings. How great is this?”
Since its establishment in 2018, the O-RAN Alliance has surged from its five founding members (all operators) to more than 260 members. Among the three major suppliers, only Huawei is not a member because it believes that the performance of the Open RAN system is not as good as the company's proprietary system. Other Open RAN teams are also growing at a similar rate. For example, the Open RAN Policy Alliance was established in May 2020, and more than 60 members are committed to coordinating global policies regarding the development and deployment of Open RAN.
Rakuten engineers can install a 4G base station for its Open RAN deployment in just 8 minutes.
In recent months, Japanese e-commerce giants Rakuten Mobile and Dish have pledged to use Open RAN for a wide range of new 5G deployments. After the British government authorized the stripping of all Huawei components from the wireless network, UK-based Vodafone is replacing these components in its network with Open RAN equivalents. Due to similar tasks, local operators in the United States, such as Inland Cellular in Idaho, are doing the same.
These deployments do not always go according to plan. Rakuten especially faces some initial setbacks when the performance of its Open RAN network does not match the performance of traditional end-to-end systems. However, operators remain optimistic and have not given up. Many people in the industry don't care about this type of problem. They think that the only way to truly eliminate technical flaws is to deploy it on a large scale and see what works and what needs improvement.
There are still lingering questions about the dollar's stop loss position. When an operator purchases an end-to-end system from Nokia, Ericsson, or Huawei, it also knows that it can rely on that supplier to support the network in the event of a problem. This is not the case with open RAN deployments, in which case no vendor may claim responsibility for interoperability issues. Larger operators may be able to support their own Open RAN networks, but smaller operators may rely on companies like Mavenir, which position themselves as system integrators. However, critics believe that this approach simply creates another end-to-end provider for operators who do not have the expertise or resources to support their own networks, and adds additional costs.
Eventually, when the next generation of wireless technology needs to be implemented, the real test of Open RAN may come. Rajat Prakash, chief engineer of Qualcomm Wireless R&D, said: "I think 6G will be built with Open RAN as a prerequisite.
It remains to be seen how far this movement will go in terms of decomposing RAN, opening up new interfaces, and even introducing new technologies into hybrids. The important thing is that the movement has gained tremendous momentum. Although some corners of the industry still have reservations, operators and small-scale suppliers have put too much weight behind the idea that this movement has failed. Open RAN will continue to exist. As it matures, the wireless industry will open the door to a new way of operating.
This article appeared in the print edition of May 2021 as "5G First Mile Conflict".