Microwave Journal

A mmWave Power Booster for Long-Reach 5G Wireless Transport

September 9, 2022

A mmWave power booster enables increased output power by more than 10 dB without appreciably deteriorating signal integrity. It enables high-power 71 to 86 GHz communications up to 10 Gbps in a 2 GHz channel and paves the way for commercial mmWave links up to 5x current link lengths at E-Band.

As mobile communications evolve into their fifth generation, the offer of higher data speed and reduced latency puts a strain on the radio access sites and on the whole network, including its transport segment.1 For wireless backhauling connections, as well as fronthauling and midhauling links as envisaged by the open-RAN paradigm,2 this entails capacities larger than 10 Gbps and advanced fast processing capabilities.

To deal with similar needs, existing spectrum regulations traditionally offer channels in the microwave range (6 to 42 GHz), but these are too narrow to support more than 1 Gbps in each; while 2 GHz channels are available in the mmWave domain, specifically in the commercial E-Band (71 to 86 GHz) where a 128-QAM modulation can easily sustain 10 Gbps. Network operators are thus planning and deploying equipment for E-Band communications, and equipment manufacturers spend substantial effort in researching and developing such products.

The shift to E-Band, however, comes with some steep costs: 1) difficulty to obtain high power from available components, 2) larger propagation attenuation and 3) higher sensitivity to precipitation (i.e. rain, hail and snow). All of these concur to limit the achievable link distance, hence the modern wireless tradeoff of capacity versus distance.

While the latter two costs express physical phenomena, the former represents a technological limitation, which can be tackled. Part of the answer is provided by the use of large, more directive antennas, in turn requiring precise compensation of misalignments.3 Instead, this paper outlines a complementary answer: a novel E-Band power booster to enable high capacity backhauling, fronthauling or midhauling in the mmWave region, which has been developed and tested in the field.


The typical architecture of commercial backhauling equipment is based on a digital modem, some analog baseband or intermediate frequency circuitry and up-conversion to 71 to 86 GHz followed by a power amplifier (PA). Current commercial equipment dedicated to E-Band backhauling offers a transmitted power between 10 and 20 dBm at the various modulations.

Figure 1

Figure 1 Current single PA architecture with gain G (a) and parallel architecture with N identical PAs fed by a driver power amplifier (b).

Figure 2

Figure 2 16-way binary division with one to eight splits in waveguide, the last using an alumina splitters.

Increasing the output power through the same components would cause severe distortion of the transmitted signal. Regardless of the traveled distance, it would reach the receiver with excessive distortion for error-free communication even after standard forward error-correcting schemes are applied.

Integrated PAs can leverage different semiconductors, chiefly GaAs, SiGe and GaN. Most of today’s RF circuitry is GaAs based. Cost and power-handling capabilities, however, are directly related when switching to low-power cheap (in volume) SiGe and high power, but expensive, GaN. A reasonable tradeoff can instead be realized with an architectural change, by employing parallelization, while still using GaAs components.

As shown in Figure 1, a driver amplifier provides, via an ideal division network, the same input power to each of the N PAs as in the single amplifier case. In the simple third order model of non-linearity, each amplifier produces a distortion component proportional to the cube of the input power. After proper in-phase recombination, the same signal-to-distortion ratio is obtained as with the single amplifier but with a much higher level of transmitted power; 10log10(N) = 12 dB is the theoretical output power increase achievable with this architecture when N=16 identical parallel PAs are used.


For a paralleled architecture to be of practical interest, the division and recombination network must avoid power loss along the path. To this purpose, the structure shown in Figure 2 includes a one to eight waveguide binary tree,4 providing low loss due to the fine surface finish of the inner metal walls. The binary tree guarantees uniformity in amplitude and phase. To minimize reflections at the interfaces and maintain isolation between the connected amplifiers, the basic node is designed as a so-called “magic tee.”5 The final further division by 2 is implemented in planar technology on alumina. The same binary tree is mirrored to serve as recombination network.

The waveguide networks are machined in two aluminum lids in which the waveguide trees and the internal loads are milled (see Figure 3). The lids close as a sandwich to the inner aluminum body, which feature small pyramids to minimize impedance mismatches and seal the carved waveguide paths. The common input and output ports, in WR12, are located on the bottom side, where the 8+8 ports toward the amplification chains are visible on the top. Each tree exhibits a net 0.7 dB loss (beyond the theoretical 9 dB) and a total phase imbalance less than 15 degrees between all ports.

Figure 3

Figure 3 Waveguide divider network shown at the top right, with the lid detached from body. The recombination network shown at the bottom left, with the lid and body assembled.

Figure 4

Figure 4 PA structure, including the WR12 waveguide launchers, alumina splitters and PA die.

The 3 dB alumina splitters used as the topmost stages introduce a further 0.5 dB of loss each along with the various bonded interconnections shown in Figure 4. All chains are fed with good in-phase and equal-amplitude signals and their recombination follows the same rule; considering recombination losses, an expected practical increase of output power of +10 dB with respect to the single amplifier case is the expected outcome.


Table 1

The entire power booster is assembled in a first version with eight PAs without the alumina splitters, and then later in its full-fledged version with 16 PAs. Table 1 summarizes the results from a 2-port characterization between 71 and 86 GHz.

These measurements show an increased variability versus frequency in the performance of the 16-PA version, due to the alumina splitting stage and the additional manual bonding introducing further uncertainty. The 16-PA version, however, exhibits an output power at 1 dB of compression (OP1dB) that is as much as 3 dB higher than the 8-PA version, thus enabling higher sustained output power.

The power booster is tested on actual modulated signals to assess whether the currently required signal-to-distortion ratio (S/D) is achievable at a +10 dBm output power level with respect to a state-of-the-art commercial unit based on a single PA (ALFOplus80HDX by SIAE MICROELETTRONICA). The ALFOplus80HDX full-outdoor unit is capable of 20 dBm in 4-QAM (2 Gbps throughput over 2 GHz bandwidth) and 13 dBm in 128- and 256-QAM (up to 10 Gbps), with a guaranteed overall S/D at the receiver of 29.5 dB in nominal conditions of received power and with a factory-calibrated transmitter.

Table 2

By adding the power booster to an ALFOplus80HDX configured as transmitter-only and using a power meter and another ALFOplus80HDX as a receiving terminal, the performance summarized in Table 2 is measured. The columns compare the transmitted power (PTx) and S/D ratio (as reported by the receiving equipment), which measures the quality of the received signal at a prescribed received power (dominated by distortion from the transmitter) in three configurations:

1. A standard ALFOplus80HDX equipment (without any power booster)

2. An ALFOplus80HDX with power booster but without any specific recalibration of predistortion coefficients

3. The same equipment with the power booster after recalibrating the predistortion coefficients of the transmitter.

Predistortion is a numerical process which allows compensation of nonlinearities introduced by the analog stages and should thus be fine-tuned for a specific transmitting chain, which justifies the large improvement of signal-to-noise ratio after recalibration. The results show that the power booster enables increased output power by more than 10 dB without appreciably deteriorating signal integrity after recalibration of the transmitter predistortion. It enables high-power 71 to 86 GHz communications up to 10 Gbps in a 2 GHz channel.


In the wake of successful laboratory tests, a field trial in cooperation with Deutsche Telekom/COSMOTE Greece uses an outdoor booster connected to existing ALFOplus80HDX equipment. However, since the booster acts only in transmission, while the equipment uses frequency-division duplexing to transmit and receive simultaneously through the same physical antenna port, the prototype includes:

1. A duplexer (to separate transmitted and received bands from the equipment)

2. The power booster on the transmit path

3. A straight waveguide on the receive path

4. Another duplexer (to expose a unique antenna port).

This waveguide structure, required only for the field prototype, is enclosed in a metal container providing heat sinks and mechanical interfaces with the equipment and the 60 cm antenna.

To compare in real time the advantages of the power booster, only one end of the trial link is equipped with the booster prototype, whereas the other end includes only an ALFOplus80HDX and a 60 cm parabolic antenna (see Figure 5). The direction from the terminal with the power booster transmits at 74 GHz, where the direction from the terminal without the booster transmits at 84 GHz, both over 2 GHz channels.

Figure 5

Figure 5 Athens link with power booster.

Figure 6

Figure 6 Athens field trial using a site equipped with the power booster (left), transmitting at 74 GHz and receiving at 84 GHz, and a site without the booster (right), transmitting at 84 GHz and receiving at 74 GHz.

Two suitable line-of-sight sites in the Athens region were identified with a separation of 4.6 km (see Figure 6). Monitoring equipment was also installed to record received power levels and modulation in both directions by querying the local and remote equipment every 2 seconds. After aligning the antennas on a clear day, received measured power levels are shown in Table 3. The values highlight an improvement of more than 10 dB in received power because of the power booster. This considers the accuracy of the internal power meter of about ±1 dB and the inherent differences between the two directions due to the slightly different transmit frequencies.

Table 3

Both ends were configured to use automatic modulation, so that the two ends automatically maintain the highest modulation compatible with error-free communication in the instantaneous link conditions (i.e. rain fading). Both equipment switched automatically to the maximum capacity of 10 Gbps.

Over the monitoring period, the link transported more than 2,000 Pbit (1 Pbit = 1015 bit) in each direction, as the maximum 10 Gbps (128-QAM modulation) could be maintained for the vast majority of time, where rain events occurred only in a few days. A sample of the monitored received power and modulation is shown in Figure 7.

Figure 7

Figure 7 Received power and modulation vs. time. The green traces show the power and modulation received from the transmitter with the power booster; the blue traces show the power and modulation received from the transmitter without the booster.

An exceptional rain event with torrential precipitations in the Athens area was recorded, with rain intensity greater than 100 mm/h, causing numerous power outages and unreachability of the network elements. Several tens of minutes of unavailability were gathered in both directions during this severe thunderstorm. Discarding this data, the rest of several seasons included light as well as medium rain events, obtaining the aggregated performance in Table 4.

Table 4

The direction that leverages the power booster reduced downtime by 91 percent and the non-maximum modulation (corresponding to less than 10 Gbps per direction) time by 70 percent with respect to the same link without the booster in transmission.


The modern needs for very-high capacity demanded by modern 5G networks is hampered by physical and technological constraints that limit the reachable hop length in wireless mmWave transport. The adoption of a parallelized PA architecture, however, circumvents some of the hurdles and so enables long-reach connections in the commercial E-Band.

The power booster prototype relies on a one-to-eight waveguide distribution and recombination structure feeding double PAs that yields a 10 dB increase in transmitted power. A 4.6 km field trial monitored through a multi-seasonal period validates the approach, thus paving the way for commercial mmWave links up to 5x current link lengths at E-Band.

Future activities will be dedicated to integrating the power booster in next-generation wireless transport equipment and industrializing the product, while also investigating alternative technologies such as GaN.


  1. A. Nordrum and K. Clark, “5G Bytes: Millimeter Waves Explained,” IEEE Spectrum, May 2017. Web. https://spectrum.ieee.org/video/telecom/wireless/5g-bytes-millimeter-waves-explained.
  2. “O-RAN Architecture Overview,” O-RAN Alliance, Web. https://docs.o-ran-sc.org/en/latest/architecture/architecture.html.
  3. M. Oldoni, S. Moscato, G. Biscevic and G. Solazzi “A Steering Antenna for Long-Reach mmWave X-Haul Links,” Microwave Journal, Vol. 21, No. 10, October 2021.
  4. S. Moscato, M. Oldoni, G. Cannone, D. Tresoldi, A. Pini and A. Colzani “8-way Paralleled Power Amplifier for mm-Wave 5G Backhauling Networks,” European Conference on Antennas and Propagation, March 2021.
  5. “TEE Junction | E-Plane Tee, H-plane Tee, Magic Tee,” Electronics Club, Web. https://electronics-club.com/tee-junction-e-plane-tee-h-plane-tee-magic-tee.