The introduction of 5G stimulates the mobile industry to apply advanced optical technologies.
Existing 2G / 3G / 4G networks are built, as a rule, on the basis of the so-called distributed architecture of a radio access network (Distributed RAN, D-RAN), in which the components of the base station are the digital processing module BBU (Baseband Unit) and the radio modules RRU (Remote Radio Unit) – located on one site. This arrangement allows you to minimize the required capacity of the transport network. Depending on the topology, the capacity of the “last mile” transport domain is from 1 to 10 Gbit/s.
However, as you know, for 5G, the centralized RAN architecture (Centralized RAN, C-RAN) is optimal, in which the computing resources (BBU modules) of the base stations are combined into hubs. C-RAN is a form of introducing centralization and virtualization solutions VNF / NFV into the radio subnet as a strategic direction in the development of telecommunications infrastructure. The benefits of VNF / NFV are well known – from lowering barriers to new manufacturers and lowering capital and operating costs to accelerating the launch of new services and network upgrades. For a radio network, these advantages of economic efficiency are enhanced by an additional increase in the bandwidth of the radio channel (centralization of radio resource planning functions, reduction of interference, etc.).
Until now, the industry focused on the core network, and now it can be stated that the problem of virtualization and centralization in this domain has been solved. However, considering that investments in the radio network with the introduction of 5G are estimated at approximately 70% of the total volume, centralization in the RAN is key to achieving the commercial attractiveness of 5G networks.
The starting point for the implementation of centralization and virtualization technologies in RAN is the new layout of the base station, in which the digital signal processing stack, previously concentrated in the BBU, is now divided into two functional units – the DU (Distributed Unit, a real-time digital radio signal processing module) and CU (Centralized Unit, a module for digital processing of slow processes). This separation provides the flexibility of choice and the effectiveness of building the RAN at each BS site, taking into account restrictions on placement, transport channel features, application conditions, etc.
Another problem solved by the new layout of the base station is related to the inefficiency of the Common Public Radio Interface (CPRI) between the RRU and the BBU, in which the required bandwidth is proportional to the radio channel width and the number of antennas (for example, already in the LTE MIMO 2×2 network bandwidth required for CPRI , 13-15 times more subscriber traffic speed). During the implementation of 5G with an increase in the width of the radio channel (when mastering high frequency ranges) and the number of antennas (when using the Massive MIMO solution), the requirements for the band in the transport network grow at a faster pace than the growth in the speed of subscriber applications. This can be a problem for the construction of the C-RAN, where exactly the data transfer rate between the RRU and the BBU will determine the capacity of the “last mile” transport domain.
Reducing bandwidth requirements in a new generation base station is achieved by transferring some of the low-level digital processing functions from BBU to RRU. And for signal transmission after intermediate digital processing in RRU, a new interface is used – eCPRI (enhanced CPRI), which features a variable, load-dependent speed and the use of a packet Ethernet transmission medium.
The new layout of the BS provides flexibility and high economic efficiency due to the centralization of computing resources and the widespread introduction of virtualization technologies in the RAN, taking into account various limiting factors, including technological ones (the delay in the transport channel for CPRI / eCPRI interfaces should not exceed 100 μs). Until recently, the transport channel requirements between the RRU and the BBU / DU / CU in the CPRI were not specified. Now this gap has been eliminated, specifications have been developed that define the quality requirements of the transport network for CPRI and eCPRI. This opens up opportunities for operators to choose the best technical solutions for different vendors, thereby avoiding the monopoly of the supplier, whose base stations are installed on the network.
The impact of RAN architecture in the implementation of 5G on the transport network
One of the key characteristics of 5G networks is a more than 10-fold increase in data transfer rates for subscribers. Therefore, even within the framework of existing architectural solutions, the requirements for the bandwidth of the transport access network also increase by more than 10 times. And during the construction of 5G networks with the target C-RAN architecture, the requirements for the transport access network increase by more than 100 times. So, in particular, it happens on a 4G / 5G network, in which existing IP / MPLS solutions extend to the delivery of traffic from the places of its generation (BS sites) to the peripheral centers of its processing.
One of the main conditions for the development of the RAN network to date has remained the “axiom of limited transport lane” and the resulting need to save this lane. For this reason, they tried to transfer the use of IP functions as close as possible to the sources of traffic generation. Following this axiom determines the extensive development of the transport network, i.e. the increase in throughput and the successive replacement of IP access nodes with more productive ones is proportional to the expansion of the mobile network, which is determined by the growth in speed and volume of traffic.
Extension of the channel capacity of the transport network for centralization and virtualization in the RAN
Transport solutions used on modern RAN networks, in terms of their bandwidth, correspond to the level of development of optical transmission technologies of the late 80s of the last century. However, today the channel capacity of optical data transmission systems is growing much (1000 times) faster than the speed in mobile networks since the construction of LTE networks (10 times). In addition, the cost of optical systems is constantly decreasing. Therefore, their use in the construction of the C-RAN transport network will remove the restrictions on the choice of the optimal RAN architecture imposed by the capabilities of traditional transport networks. The family of such alternative solutions for building a C-RAN network was named in the industry by the name of the corresponding transport domain – X-Haul
The advantages of the traditional IP / MPLS approach are manifested in the intensive use of its basic functions, i.e. aggregation, routing, traffic transit between nodes in a network of complex topology. With centralization (enlargement), the number of nodes decreases, and therefore, the number of routes decreases and the topology is simplified. In an IT network, all data centers are usually connected by point-to-point channels.
C-RAN / vRAN – a convergence zone with the IT industry in the field of optical solutions
With the beginning of the widespread use of virtualization and cloud computing technologies, the initiative for the development of optical systems moved to the IT field, where the key requirements are to organize high-speed channels over short distances (<300 m, <2 km) inside or between data centers (<10 km, <40 km, <80 km) with the highest density of ports (which implies microminiaturization and reduced energy consumption) at low cost and large production volumes. This is fully consistent with the requirements of the mobile business. The economy of the latter relies on a reduction in the cost of network infrastructure, which is achieved through the unification and standardization of the solutions used, their mass production and the high capacity of the global market.
At the moment, the development of the optical industry mainly follows the development of semiconductor technologies for the production of optical systems and the introduction of modulation methods with a low implementation cost.
Integrated Silicon Photonics
By analogy with the transition from electronic circuits on distributed elements to integrated circuits, the transition of optical networks to the production of solid-state photonics due to the placement of all elements on a single chip implies a rejection of the traditional approach involving assembly and individual precision alignment of components.
A significant economic advantage of such a transition is the ability to use the same semiconductor materials, microchip design methods, manufacturing techniques, etc. This use of investments already made in the manufacturing capacities of the manufacture of electrical microcircuits can dramatically reduce the cost of optical and optoelectronic systems. The use of semiconductor technologies for manufacturing optical systems breaks the relationship between their complexity (number of elements) and transmission speed, on the one hand, and their cost, on the other, with a dramatic increase in reliability. Therefore, when using solid-state photonics, you can get a transport solution that meets the increasing requirements for the bandwidth of the transport channel for C-RAN, without increasing its cost.
To date, the production volume of integrated optical modules already exceeds the volume of optical modules developed and made of discrete components, and the growth of the market, determined by the development and needs of data centers, leads to a drop in their cost by 10–20% annually with a multiple reduction in size and power consumption.
Today, commercial products use various technological approaches to optoelectric integration: it can be performed on a single monolithic crystal or on different crystals (the so-called hybrid integration). The advantages of the latter are the ability to choose the optimal solution for each of the segments (electrical and optical). Thus, the manufacture of electrical microcircuits, as a rule, requires the use of photolithography technology with a higher resolution than for optical microcircuits.
Cost-effective modulation techniques
The choice of one or another modulation for the implementation of the optical system is determined by the conditions of its application. For the telecommunications market, optical solutions are usually created on the basis of coherent modulation and include amplitude-phase modulation methods with a large number of encoded bits per symbol (QPSK, 8QAM, 16QAM …), polarization multiplexing, laser on the receiving side for restoration signal phases, as well as a digital signal processor (DSP) for analog-to-digital signal conversion with compensation for various types of dispersion, etc. Such optical systems are often characterized by a lack of optical compatibility between products from different manufacturers and sufficiently large sizes (form factors CFP, CFP2) due to power consumption and the need to accommodate a large number of components.
In optical systems for the IT industry, which are developed for large-scale production and application over short distances, the most important criterion is to minimize the cost of the optical module or transmitted bit. And here the main approach is to use the “base speed” with simple types of direct detection modulation (IM-DD, Intensity Modulation – Direct Detection) with a widely available component base, low cost and power consumption (NRZ, PAM4, etc.). A further increase in speed is achieved by increasing the number of carriers – by multiplying the “base speed”.
It is important to note that in the mobile industry, CPRI and eCPRI interfaces are most often used for speeds up to 25 Gbit / s, i.e. values of the “base speed” of the IT industry. In addition, the planned optical channel length for X-Haul solutions coincides with the most common optical path length for data exchange between data centers.
Thus, the construction of a 5G RAN radio access network based on optical channel transmission systems will allow operators not only to obtain the expected benefits of centralization and virtualization, but also provide the most efficient transport solution due to convergence with the IT industry in the field of optical systems.