20 Mar Photonic band gap fibers could help mitigate the latency challenge in 5G transport networks
There is more that distinguishes 5G from 4G LTE and other earlier generations of mobile networks than the promised ultra-high speed that most of us are excited about. Low latency and improved Quality of Service (QoS) are even more important attributes, especially for the Ultra Reliable and Low Latency Communications (URLL) use case. The importance of latency is evidenced by the plethora of research and development activities focused on reducing latency in all aspects of 5G networking – air interface, network architecture, edge computing, frequency multiplexing, beam management, traffic scheduling, packet length etc. This article focuses on the fiber optic transport network contribution to latency and introduces hollow core photonic band gap fibers and their potential role in improving overall end to end latency and provide more network design flexibility for operators.
Before now, the evolution of mobile networks primarily focused on improving speed and meet demand for high bandwidth applications such as high-definition mobile TV and video conferencing. Migration to 5G represents a paradigm shift in mobile networking that does not only focus on high bandwidth applications but also on latency sensitive and machine type applications. The 3rd Generation Partnership Project (3GPP), a standards organization for wireless networking, has defined three use cases for 5G:
- eMBB (Enhanced Mobile Broadband) for high speed applications. Such applications can be achieved without significant changes to the 4G LTE infrastructure
- mMTC (Massive Machine Type Communications) for high reliability narrow band Internet access for applications such as smart metering and power monitoring and control
- URLLC (Ultra-reliable low latency communication) for low latency applications such as remote surgery and autonomous driving. While full level 5 autonomous vehicles rely on on-board vision systems to navigate their way and Wi-Fi like broadcasting systems to communicate with other vehicles, they will also rely on 5G networks. 5G will enable them to access, in real time, information critical for safe driving such as high definition 3D maps and to achieve one on one smart vehicle communications.
To achieve these diverse applications which have stringent specifications, 5G networks will use the so- called network slicing. This is the implementation of multiple virtual networks over a single infrastructure using software defined networking (SDN) and network function virtualization (NFV). Each virtual network will address a different application space. But virtualization is only as good as the underlying physical network which must be designed to overcome all the relevant network challenges, such as latency.
Evolution of mobile transport networks
Prior to 4G LTE, mobile radio access network (RAN) transport consisted of two segments, the fronthaul and backhaul. The fronthaul, typically co-axial copper, connected the remote radio head (RRH), mounted near the mobile antenna, to the baseband unit (BBU) at the base of the tower. The backhaul, usually a microwave link or fiber, connected the BBU to the mobile core network. As the industry migrated to 4G networks to meet the higher bandwidth demand and to cater to increasing mobile subscribers, there was too much pressure on mobile base stations. The traditional role of the base station had to be divided into two functional units, the radio unit (RU) and a central unit (CU) in the so called centralized or cloud radio access network (C-RAN). The central unit was moved up to 20km away from the base and could process traffic from multiple towers. The distance and bandwidth involved required optical fiber fronthaul to connect the RU to the CU. For most mobile networks, the use of optical fiber in the backhaul, connecting the CU to the core, became inevitable.
Requirements for 5G have become even more complex and need a brand-new approach to RANs. 3GPP has introduced new interfaces and modules for 5G. In addition to the RU and CU, a distributed unit (DU) and other modules and concepts (beyond the scope of this discussion) have been introduced in a new RAN, dabbed NG-RAN (or next generation RAN). The 5G RAN offers multiple architectural options to operators, such as how to split the RAN functions, how they should be placed in the network and the transport links required to connect them.
The latency challenge in 5G RAN for URLLC
The above figure is a schematic of a 5G RAN layout showing functional modules and how they may be connected by optical transport links – the fronthaul, midhaul and backhaul. It also shows the range of latency budgets assigned to each segment and the range of possible distances than can be adopted.
- The fronthaul connects the RU to the DU
- The midhaul connects the DU to CU
- The backhaul connects the CU to the core network
The onus is on the mobile operator to determine the location of the functional modules that will best meet their customer needs while complying with the latency recommendations. If the application is eMBB, then it is relatively easy to meet the maximum distances while complying with the latency guidelines and meeting the high bandwidth specification. It is more than likely that all operators who have already announced 5G availability have only achieved the high bandwidth requirements of 5G or the eMBB use case.
Meeting the URLLC requirements is more tedious and requires two-way end to end latency of no more than 1ms. This stringent budget includes not only fiber optic transport but also air transmission (from user device to antenna), signal processing time and several other latency sources. Moreover, URLLC includes a high reliability requirement that calls for very stringent block error rate (BER) of 10-9 to 10-5, a drastic change from the 10-2 required for 4G LTE. This is a competing requirement to latency because low BER requires high signal to noise ratio (SNR) that can be achieved, in part, by time consuming signal processing. Block error rate is the wireless transmission equivalent of bit error rate and is defined as the ratio of the number erroneous blocks to the total number of blocks transmitted.
Recall that the two way latency (or group delay) of optical signals propagating through an optical fiber is 2L/vg or 2Lng/C, where L is the length of the fiber, ng (=1.482 in standard single mode fiber at 1550nm) is the group index of refraction, C(~3×108 m/s) is the speed of light in a vacuum. With these assumptions, 1ms of latency translates into 100km of standard single mode fiber. But since the 1ms latency budget is shared by other modules and processes, the RAN must be designed to cover a distance far less than 100km. Different network considerations such as the use of edge computing servers closer to antennas or collocating two or more functional modules have been adapted to meet the distance constraint. One proposal for autonomous vehicles is to collocate edge computing servers with electric charge stations along the streets, close enough to the RAN functional modules.
Any reduction in fiber latency by fiber optic manufacturers will offer some network design flexibility to operators and service providers. Low latency hollow core photonic bandgap fibers can go a long way to address the distance constraint imposed by the low latency requirement.
Hollow core photonic bandgap fibers
Unlike commonly used optical fibers based on total internal reflection, the propagation of light in photonic bandgap fibers is based on the photonic bandgap effect – the unusual interaction of light with photonic crystals. A photonic crystal (or photonic bandgap material) is a periodic structure of materials with different indices of refraction which can prohibit photons with a certain energy range to go through the material. Since photon energy is directly proportional to the frequency of the light or inversely proportional to the wavelength, then the photonic bandgap material will prohibit light with a certain wavelength range to pass through it. The choice of photonic bandgap material, the periodicity of the structure and other factors determine the location of the bandgap on the electromagnetic spectrum and the width of the bandgap. Manufacturers can thus fabricate photonic crystals for specific bandgaps for different applications by manipulating these parameters.
In hollow core photonic bandgap fibers, the fiber core is an air hole while the cladding represents a photonic bandgap region of alternating air holes and silica. If light whose wavelength matches the bandgap is launched, it is guided through the hollow core and is “forbidden” from escaping into the cladding. The refractive index of the air core region is almost 1 and can significantly improve the latency of the fiber by about 30% over standard single mode fiber. In addition to lower latency, the hollow core of such fibers is significantly less prone to nonlinear impairments than their conventional counterparts. Consequently, more power can be transmitted through the fiber and improve optical signal to noise ratio (OSNR),
Hollow core fibers have long since been commercialized for all the communication spectral windows. Most commercial HC-PBG fibers target specialty applications such as fiber optic gyroscopes, pulsed lasers and gas spectroscopy. For these fibers to be more widely adapted in communication applications, more work is required to bring down the attenuation, currently in the 10s of dB/km, in line with conventional step index fibers. Laboratory experiments have demonstrated the feasibility of lowering the attenuation below 1dB/km. With the huge latency challenge in 5G transport networks, there is enough motivation for manufacturers to develop and manufacture lower attenuation hollow-core photonic bandgap fibers.
5G mobile networks represent a revolution in wireless networking and will enable applications that, until recently, only sounded futuristic. But before all the 5G enabled possibilities can be realized, a myriad of technical challenges, including latency, must be resolved. The use of hollow core photonic bandgap fibers is a viable option for mitigating latency in 5G RAN. Deployment of these fibers in 5G transport networks is a great opportunity that should motivate manufacturers to solve outstanding technical issues such as lowering attenuation.