[Introduction]5G deployment is mainly driven by two key factors, which are often contradictory: one is system capacity (spectral efficiency), and the other is system cost (energy efficiency). Spectral efficiency describes how much capacity can be provided, usually measured in bps/HZ (bits per second per Hertz), while energy efficiency describes the cost of operating a network at a given capacity.
For mobile technologies of the past, the cost has increased almost proportionally with the increase in capacity, because providing higher capacity means building more base stations or increasing the spectrum bandwidth within the network. While this approach has been maintained in the past, it will be difficult to follow if the demand for 4G network capacity increases by a factor of 10 to 100, as consumers are unlikely to be willing to pay the accompanying increase. As shown in Figure 1, in order to promote the development of mobile networks, the industry needs to solve the problem of how to reduce network operation costs while increasing the entire network capacity.
Figure 1: The 5G business case
How much does a cellular network cost to run?
Although base station energy efficiency improved significantly during the transition from 2G to 4G, the accompanying increased capacity also resulted in significant cost increases due to network densification (Figure 2). In the construction and operation of a cellular network, the vast majority of expenditures lie in the provision of air-conditioning remote control and space rental for base stations (References 1 and 2). From the initial capital expenditure (CAPEX) point of view, the cost of air conditioning accounts for more than 50%, and the rest is mainly the cost of base station equipment. Similarly, in terms of recurring operating expenses (OPEX), electricity also accounts for almost 50% of spending. Most of the power is used to run the remote distributed air-conditioning network to cool the baseband processors (the radio units are usually air-cooled and no additional air-conditioning system is required), yet the actual transmitted energy is only 7% of the OPEX. If more base stations are to be deployed, the cost of site leasing, which originally accounted for 30%, will also increase. As a result, the simple solution of deploying more base stations is not ideal (this is a big problem for 5G FR2). problem, because the cell range of the former will be greatly reduced compared to 5G FR1).
Figure 2: Power consumption of cellular networks
It is evident from the power consumption analysis that most of the expenditure comes from the distributed deployment of the baseband processing part in the base station and the remote deployment of air conditioners. China Mobile proposes to centrally deploy baseband processing in a similar fashion to Internet data storage facilities. Figure 3 shows the cloud architecture of the baseband system, that is, each baseband in the base station becomes a virtual machine in the cloud (C-RAN), and even traditional stand-alone network devices (such as gateways) can be integrated into the cloud as virtual machines. By centrally deploying baseband processing, it is possible to centralize remote air conditioning control, thereby significantly reducing OPEX and CAPEX. In addition, when decentralized base stations transmit to mobile phones with centralized control (network MIMO), CoMP is easier to implement and spectral efficiency is also improved. This system architecture is independent of the Radio Access Network (RAN) and can be used to control hybrid cellular networks.
Figure 3: Centralized baseband processing
The type of information the network transmits also has an impact on energy efficiency. As shown in Figure 2, different types of data have different packet-to-signaling packet ratios (DSR). A low DSR represents low utilization of the data transmission channel; for example, text messages, which account for 60% of all network traffic, have a DSR between 1 and 3, while photos and videos require fewer signaling packets, so energy higher efficiency. 5G FR1 addresses this challenge by adjusting the subcarrier spacing so that different types of data can use the available channel capacity more efficiently.
Determinants and expansion methods of network capacity
At the beginning of the 20th century, two researchers separately derived a relatively simple formula called Moore’s Law in the communications industry, the Shannon-Hartley Theorem. This theorem gives an upper limit on the amount of information that can be transmitted on a wireless channel, where the capacity of a single channel depends only on two parameters: channel bandwidth (BW) and signal-to-noise ratio (SNR). Although capacity scales linearly with channel bandwidth, it is only log2 proportional to signal-to-noise ratio:
According to the Shannon-Hartley theorem, there are four basic ways to increase network capacity (Figure 4):
Increased channel bandwidth: Carrier aggregation was used in 4G to increase the available signal bandwidth, while 5G FR2 uses mmWave frequencies for greater capacity.
Increase the number of channels: MIMO utilizes multipath scattering within the network to transmit on multiple channels simultaneously. Similar to channel bandwidth, network capacity also scales linearly with this effect, but the upper limit is limited by the multipath correlation (or similarity) within the network. 5G FR1 increases data rates with the optimization of MIMO.
Increased network output power: This approach has limitations due to noise in the SNR, close logarithmic scale of the SNR, and health/safety concerns involving high electromagnetic energy. Using Femtocells in areas with low coverage is one of the safer ways to improve the SNR of the entire network. However, if too many omnidirectional antenna femtocells are deployed in the same area, there will be interference between femtocells, which undoubtedly brings an upper limit to the network capacity gain. However, the energy efficiency of the network can be improved if the energy can be directed, a method known as “beamforming” (a key technology in 5G FR1 and FR2 base stations).
Figure 4: Spectral Efficiency
Improving Energy Efficiency Using Beamforming
In traditional cellular networks, a base station associated with a cell transmits energy over a fairly wide area (usually an arc of 120 degrees in front of the base station). Some of this energy is received by users in the base station cell, but most of the energy is absorbed by the environment (buildings, pedestrians, trees, cars, etc.). These losses imply a decrease in energy efficiency and an increase in network OPEX (Figure 5). If a single base station antenna is replaced with 120 antennas that direct energy to each user, the power consumption required by the base station is reduced to 0.1% of the original output power (Reference 3). However, this reduction is only a theoretical value. From a practical point of view, due to the efficiency and loss of the radio frequency components inside the base station, the output power under the same capacity can only be reduced to 30% of the original power.
Figure 5: Beamforming and Energy Efficiency
To achieve beamforming, a set of antennas at a specified interval can form a beam in any direction by simply changing the phase difference between the antennas (Figure 6). The most typical antenna array spacing is half a wavelength, so that the beam angle () is directly related to the phase difference between the antennas: . Although beamforming can focus energy in a given direction, However, it is unavoidable that energy will be transmitted to other directions (side lobes and back lobes). This extra energy will cause interference to other users in the base station cell. This effect can be mitigated by ensuring that neighboring users are in phase zero of the main beam, or by assigning weights to individual antennas through the amplitude distribution, thereby reducing the energy in the side lobes (Figure 6).
Figure 6: Principle of beamforming
There are three architectural types for beamforming that directly affect base station energy efficiency and terminals (Figure 7):
Analog Beamforming (ABF): The traditional approach to beamforming is to use attenuators and phase shifters as part of an analog RF circuit, where a single data stream is split into different paths. The advantage of this approach is that only one RF link (PA, LNA, filter, switch/circulator) is required, and the disadvantage is that the cascaded phase shifters will incur losses at high power.
Digital Beamforming (DBF): Digital beamforming assumes that each antenna element has a separate RF link. The beams are then “shaped” in a matrix-like operation, ie the amplitude and phase are manually weighted in baseband. Since RF link components are relatively inexpensive and can combine MIMO and beamforming into a single array, this approach is often the best choice for frequencies below 7 GHz in 5G FR1. For frequencies of 28 GHz and above, the PAs and ADCs of standard CMOS components are very susceptible to loss, while the use of rare materials such as gallium arsenide and gallium nitrate can reduce losses, but at a high cost.
Hybrid Beamforming (HBF): Hybrid beamforming combines digital beamforming with analog beamforming, ensuring flexibility for beamforming and multiple radios, while reducing beamforming unit (BFU) expenditure and loss. Each data stream has its own independent analog BFU and a set of M antennas. If there are N data streams, there are NxM antennas. The analog BFU loss caused by the phase shifter is mitigated by using an optional beamformer such as a Butler matrix in place of the adaptive phase shifter. The proposed architecture is to use a digital BFU to control the direction of the main beam, and an analog BFU to control the beam within the digital envelope.
Figure 7: Beamforming Architecture
Ideal Networks: Spectral Efficiency and Energy Efficiency
The combination of C-RAN, MIMO, new spectrum and beamforming will allow 5G to expand capacity while reducing costs compared to traditional and incumbent cellular networks. The Shannon-Hartley theorem can be optimized to take into account the energy efficiency of the channel (Reference 2). According to the constraints of base station and network performance, the ideal spectral energy efficiency of the combined 2G and 4G networks can be calculated, which is 4 bps/Hz for GSM and 8 bps/Hz for LTE. (It should be noted that in the real network environment, the spectral energy efficiency of LTE is often lower, generally at 4 bps/Hz).
Compared with LTE networks, in 5G FR1, which combines MIMO and digital beamforming, the capacity can be increased by more than 3 times, and the cost can be reduced by 10 times (assuming that each user corresponds to 8 transceivers with beamforming capabilities). 5G FR1 has limited spectrum available, while 5G FR2 uses a lot of spectrum above 24 GHz. The spectral efficiency of 5G FR2 (assuming a hybrid beamforming configuration of 8 transceivers per antenna array) is comparable to LTE at 10 bps/Hz, but more energy efficient (Reference 4).
Figure 8: Network Optimization: Spectral Efficiency and Energy Efficiency
To sum up, the combination of spectral efficiency and energy efficiency enables operators to both increase capacity and reduce OPEX when deploying new networks. In the future, different solutions for FR1 and FR2 will be able to be integrated into a single network, with FR1 providing high rates with building penetration in the WAN, while FR2 is used for data offloading, hotspots and extreme network densities. This network deployment will not only affect consumers and equipment suppliers, but will have a decisive impact on the entire test and measurement (T&M) industry.
Impact on the test and measurement industry
The need for new base stations in 5G has given rise to a new measurement paradigm in which both antennas and transceivers are tested over the air (OTA).
5G Base Station Architecture
Combining beamforming and MIMO into a single array results in a massive MIMO base station because both beamforming (requires the same data vector for each antenna) and MIMO (requires a different data vector for each set of beamforming antennas) Multiple sets of antennas are required. Designing base stations that can improve both spectral and energy efficiency is complex and requires very tight integration of all components (Figure 9):
Beamforming Architecture: Depends on the availability of components in terms of both loss (energy efficiency) and cost.
Broadband Power Amplifiers and Filter Banks: As the number of frequency bands increases, carrier aggregation over broadband will require a large number of filters and power amplifiers. Power amplifiers will require predistortion or rare materials to work efficiently.
Antenna Mutual Coupling: Simply putting more antennas in the space reduces base station capacity and increases losses.
Clock synchronization: For large MIMO arrays, the clocks on each PCB need to be synchronized. Clock drift causes indeterminate phase changes between antennas (due to frequency drift) and affects beamforming effectiveness.
Adaptive calibration: Due to the large number of components, chipsets, clocks, and amplifiers, coupled with the phase dependence on temperature conditions within the base station, the output phase of each antenna can deviate significantly from the desired value. Therefore, an adaptive calibration circuit measures the phase and amplitude offset of each signal, followed by predistortion to achieve excellent beamforming.
Fiber Optic Transceivers: In general, the output of a massive MIMO base station is baseband data, which is transmitted over fiber to the local baseband unit or into the C-RAN. Therefore, a real-time Field Programmable Gate Array (FPGA) is required to translate the baseband data output by the RFIC into a baseband protocol for optical fibers.
Heat Dissipation: Integrating up to hundreds of antennas, thousands of components, and dozens of RFICs/FPGAs in a confined space can cause severe thermal and high temperature problems. Since these units are deployed in areas with large temperature differences, large heat sinks are required if external air cooling is not provided, which adds significant weight to the Massive MIMO units.
Figure 9: Massive MIMO Architecture
Test Measurements for 5G Base Stations and Equipment
Traditionally, the performance of the base station is the performance of the radio frequency transceiver excluding the antenna. The performance of the RF transceiver can be directly measured by connecting the RF test port to the measurement instruments (ie vector signal analyzers and signal generators). Antenna performance is usually measured in an OTA (CW wave) manner using a vector network analyzer.
Since massive MIMO base stations are highly integrated architectures, direct access to individual RF paths is no longer possible. This means a substantial change in the way measurements are made, moving from highly predictable conducted measurements of RF transceivers to uncertain OTA measurements (Figure 10).
Figure 10: A new measurement paradigm for 5G
OTA measurements are significantly more complex than cable measurements due to the different physics of the radiated fields (Figure 11) in the near-field and far-field regions of the device under test (DUT). Due to the time- and space-variant characteristics of the modulated signal, the measurement must be performed in the far field (plane wave) of the DUT, resulting in the use of only huge antenna anechoic chambers, or indirect far-field anechoic chambers such as plane wave converters (PWC) or Crunch Field (CATR). CATR uses a reflector to convert spherical waves to a plane wave distribution in the near field of the reflector, while PWC uses an array antenna to generate a plane wave distribution in the near field (Figure 12).
Figure 11: Antenna Electromagnetic Field
Figure 12: Plane wave converter and CATR
With the elimination of RF test ports and the use of millimeter-band frequencies, OTA is expected to become an important tool for testing base station performance, not only for massive active MIMO array antennas, but also for internal RF transceivers. For these reasons, the demand for OTA chambers and measurement equipment will explode, not only for demanding measurements of antenna radiation characteristics, but also to replace traditional conducted RF transceiver measurements. Rohde & Schwarz has extensive expertise in the fields of anechoic chambers and measuring equipment. Rohde & Schwarz is fully prepared to provide complete solutions at any time in order to meet the future needs of customers (Ref. 5).
1. CMRI, “C-RAN: The Road Towards Green RAN,” Dec. 2013
2. I Chih Lin, C. Rowell, et al, “Towards Green and Soft: A 5G Perspective”, IEEE Communications Magazine, Feb 2014
3. F. Rusek, et al, “Scaling Up MIMO: Opportunities and Challenges with Very Large Arrays”, IEEE Signal Processing Magazine, Jan 2013
4. H. Shuangfeng, et al, “Large Scale Antenna Systems with Hybrid Analog and Digital Beamforming for Millimeter Wave 5G”, IEEE Communications Magazine, Jan 2015
5. Antenna Array Testing White Paper: 1MA286, 2016