A Discussion on Optical Transceivers for 5G Fronthaul Connectivity

Nach seiner offiziellen Einführung gewinnt 5G weltweit schnell an Fahrt, insbesondere in Europa, Nordamerika und Asien. Verschiedene Netzgremien haben für die kommenden Jahre ein schnelles Wachstum der 5G-Verbindungen vorausgesagt. Einem Bericht zufolge werden zwischen 2020 und 2025 Investitionen in Höhe von rund 1,1 Billionen US-Dollar erwartet, von denen etwa 80 % allein auf den 5G-CAPEX entfallen werden (siehe folgende Abbildung).

Wireless fronthaul interface of 5G networks In contrast, 5G wireless communications require more spectrum resources than 4G because they provide various bandwidth-intensive services such as;

- Ultra-reliable low-latency communications (URLLC)
- Massive machine-triggered communications (mMTC)
- Enhanced mobile broadband (eMBB).

The current 5G version relies on FR1 spectrum in the sub-6 GHz range, which offers a maximum bandwidth of 100 Mb/second, five times that of 4G LTE. With 64 available channels and 100 MHz of bandwidth, the CPRI requires at least 100 Gb/second for the fronthaul channels. However, in 2017, the industry was not yet prepared for 100 Gb/second optical transceivers, which was an important reason for the development of the eCPRI protocol.

eCPRI Split Modes - Explained

For 5G communications, optical module requirements have evolved from 10Gb/s to 25Gb/s.

Why are more optical modules needed for 5G?

In view of the overcrowding in the low and medium frequency bands, the3GPP allocated a higher frequency band for 5G. However, the higher frequency band is accompanied by higher signal loss. Therefore, a higher density of 5G base stations is required for reliable communication, and therefore a higher number of optical modules is needed for 5G networks. In other words, more optical modules are needed for a 5G network designed for a certain number of connected devices than for the same number of devices connected to a 4G network. According to market research, 50% of all optical modules sold in the next five years will be 25G for 5G fronthaul connections.

25G optical modules are mainly used to build 5G wireless fronthaul links. Thus, the telecom sector can save costs by reusing existing 25G resources.

5G Wireless Fronthaul Typical Scenarios

Centralized RAN (CRAN) and distributed RAN (DRAN) are the two typical architectures for 5G wireless fronthaul. In CRAN, BBUs are deployed in a centralized location, and therefore CRAN offers cost savings in terms of power consumption, space requirements, and operational costs. In addition, centralized baseband units or BBUs create a BBU pool that network operators can manage centrally. For large-scale 5G deployments, the CRAN architecture is preferred because the cost of building a 5G network is much higher than the cost of building a 4G network and site acquisition is another major challenge.

Direct fiber links between AAUs and DUs are suitable for both CRAN and DRAN because they are easy to maintain and cost less. In the DRAN front haul, the DUs and AAUs are installed under and on the tower, respectively, so that the distance between the two remains within 300 meters. With CRAN, on the other hand, the maximum possible distance between the AAUs and DUs can be up to 10 kilometers. For 5G optical fronthaul links, 25G gray optical modules are required. The following figure shows the difference between C-RAN and D-RAN architectures;

Direct fiber optic connections for CRAN

Fiber links in the CRAN architecture require more optical cables and transceivers. In cases where fiber resources are limited, bidirectional gray 10-km transceiver modules are more suitable because they require fewer fibers. Another way to reduce the number of fibers per 5G fronthaul link is WDM. Semi-active and passive WDM devices with colored 25G optical modules are used for WDM links.

Chinese market scenario

For a 5G macro base station, three 25 Gb/s eCPRI links are needed to cover the entire 100 MHz spectrum. In the case of China, China Uniform and China Telecom share 200 MHz of 5G spectrum, while China Mobile has 160 MHz of spectrum. In these cases, the number of interfaces increases from 3 to 6, provided the rate remains at 25 Gb/second. Each macro base station needs six pairs or twelve 25G optical modules to meet the interface transmission requirements.

Thus, a single set of 12-wavelength colored optical modules (with one fiber per macro base station) or two pairs of 6-wavelength colored optical modules (with two fibers per macro base station) can meet the connectivity needs of six interfaces.

In summary, both CRAN and DRAN scenarios will drive up the demand for 5G optical fronthaul modules.

Types of optical modules for 5G fronthaul

5G was officially released in 2019; network operators quickly began converting to this technology for commercial use. By the end of 2020, millions of 5G base stations had been built worldwide. To keep construction costs for 5G base stations in check, operators prefer colored optical modules. Based on various available WDM standards, different organizations have proposed LAN WDM (LWDM), coarse WDM (CWDM), micro-optical WDM (MWDM) and dense WDM (DWDM).

25G gray optical transceiver (SFP28 SR/LR/BiDi)

- 10-km BiDi modules (bidirectional) use DFB lasers.
- 10-km LR modules use the DFB laser type for 1310 nm wavelength operation.
- 300m range SR modules VCSEL type laser with 850nm wavelength to work.

The commercial chips required for these wavelengths are already available on the market and can be easily acquired. Some suppliers also offer industrial chips for wireless fronthaul applications.

25G colored optical modules (SFP28 MWDM/CWDM/DWDM/LWDM)

Currently, a lot of research and development work is being carried out to develop various 25G colored optical modules. These modules are developed based on existing and original WDM standards.

CWDM optical modules are installed directly on AAUs and DUs, and external multiplexers and de-multiplexers are used. ITU-T G.694.2 defines the CWDM standard, which provides six wavelengths spaced at 20 nanometers. In the 5G wireless fronthaul configuration with three channels, six wavelengths are required. Preferably, CWDM6 should include wavelengths of 1271, 1291, 1311, 1331, 1351 and 1371 nanometers because the first four wavelengths are similar to the first four wavelengths of CWDM4. Thus, chip manufacturers need to accommodate only the last two wavelengths. In comparison, 12 wavelengths are needed in six-channel scenarios, for which we can opt for 2 x CWDM6 and two fibers or 1 x CWDM12 and one fiber. CWDM12 is possible by adding six wavelengths of 1471, 1491, 1511, 1551 and 1571 nanometers in CWDM6.

MWDM

MWDM was proposed by the China Communications Standards Association in late 2019. In MWDM, 12 wavelengths are obtained by extending six wavelengths of CWDM6 through a thermoelectric cooler. The 12 MWDM wavelengths are unequally spaced and include 1267.5, 1274.5, 1287.5, 1294.5, 1307.5, 1314.5, 1327.5, 1334.5, 1347.5, 1354.5, 1367.5 and 1374.5 nanometer wavelengths. Compared to the CWDM6, the MWDM12 is equipped with an additional thermoelectric cooler (TEC) and its driver.

LWDM

With LWDM technology, the channel spacing is 800 GHz (or about 4.4 nanometers). This allows more wavelengths to be offered in the O-band at the cost of a slight degradation in dispersion. The 400GBASE-LR8 interface is defined by IEEE 802.3 based on LWDM8 at 1273.54, 1277.89, 1282.26, 1286.66, 1295.56 and 1300.05, 1305.58 and 1309.14 nanometers. The last four of these wavelengths are common for 100GBASE-LR4. The CCSA has integrated four wavelengths (1269.23, 1291.10, 1313.73, and 1318.35 nanometers) to the above eight LWDM8 wavelengths to create LWDM12. The optical chip is the only difference between MWDM12 and LWDM12.

DWDM

ITU-T G.698.4 forms the basis for DWDM technology and is widely used in metro and backbone networks. With DWDM, wavelengths range from 1520 nanometers to 1567 nanometers, spaced approximately 0.78 nanometers apart. The wavelengths supported by this technology can be 96, 48, 40, 20, 12, or 6. However, DWDM is an expensive choice due to higher transceiver costs and is considered in scenarios with insufficient fiber resources.

Welche ist die beste Wahl?

As mentioned earlier, MWDM requires TEC controllers and custom wavelength chips because of the narrow wavelength spacing. For LWDM, on the other hand, the supply chain for DML optical chips is not yet mature, and the cost of the EML laser is high. For DWDM, the chips used are expensive, and TEC controllers are also needed. Therefore, CWDM6 is the only available option that does not require TEC controllers, and the DML lasers are readily available in the market. For this reason, CWDM6 is considered the most cost-effective and viable solution for network operators.

Wireless fronthaul transceiver limitation

The transmission distance of standard wireless optical fronthaul modules is limited to 10 kilometers. However, operators will require a greater transmission distance if they choose to deploy CRAN. According to research, 3% of all deployed gray optical modules will require a transmission distance greater than 10 kilometers in the next five years. The following research-based diagram shows what experts expect for the coming years!

100G DSFP Optical Transceiver Modules

It is quite obvious that we will need more fronthaul transmission capacity as 5G develops. However, the baseband cards in cellular base stations are equipped with fixed panel ports. Therefore, equipment manufacturers need to think about ways to improve the transmission capacity of the ports.

DSFP optical transceivers are an excellent solution for the growing 5G fronthaul requirements. Although mainly used for Ethernet protocols, DSFP can be used for various wireless eCPRI fronthaul configurations. The DSFP standard was released in 2018 and offers a maximum transmission rate of 100Gb/s. DSFP modules are compatible with the structure of SFP modules. The DSFP module's built-in encapsulation tendency allows it to operate with two channels for transmitting and receiving data, doubling the transmission capacity. Currently, 25G SFP28 transceivers are the standard.

25G SFP28 tunable color optical modules

C-RAN plays an important role in the deployment of the 5G network infrastructure. In the future, we will need more colored transceiver modules. Initially, optical CWDM6 transceivers are widely used due to their easy availability and low price. However, wavelength configuration is time-consuming and labor-intensive, so color tunable DWDM technology has been proposed.

The tunable and fixed DWDM wavelength systems have similar wavelength ranges and spacing. The only difference between the two is that tunable DWDM modules can support automatic configuration from 12 to 40 wavelengths. At this point, it is important to mention that tunable DWDM is not a new concept, as it has already been used in the transport network. However, it is much more expensive than CWDM6 and efforts are being made to reduce the cost.

Conclusion:

The eCPRI standard defines the characteristics of 5G fronthaul interfaces. 25G fronthaul interfaces are fully compliant with Ethernet protocols. Therefore, existing 25G Ethernet optical modules can also be used to build 5G fronthaul interfaces. Operators are now looking for more cost-effective solutions to cover the rising costs of 5G base stations. Color tunable 25G transceivers have proven to be a practical and viable solution in the current scenario where fiber resources are limited and more expensive.