While 5G wireless standards are still under development, it’s not too early to predict that 5G device designs will be more complex, have more components (particularly filters), and be expected to deliver higher networking and processing performance. At the same time, they will be smaller and less expensive. The 5G networking standards now under development intend to accommodate a wide variety of use cases that are now served with disparate technologies. These range from low-bandwidth Internet of Things (IoT) to high-bandwidth video.
The challenges that come with accommodating these use cases will impact every part of 5G deployments, but perhaps will add the most complexity and challenge to the RF front-end (RFFE) in mobile devices—if for no other reason than there is very little space to accommodate this complexity. A full understanding of the impact of 5G networks on RFFE starts with the network environment in which the devices will work.
The 5G radio access network (RAN) is expected to be a combination of technologies, nodes, and frequencies, and this mix will result in one of the biggest challenges for 5G deployment. Densification of the network will require:
New models that make deployment economically viable. Planning for 5G deployment will be extremely difficult, and the large number of nodes will make non-optimum sites the norm. Increasingly, deployments will include shared equipment between carriers.
Dynamic and adaptable allocation of resources to maximize performance, increasing automated software control. This will include interference coordination and capacity allocation, even in unplanned and chaotic environments. Coordinated multipoint (CoMP) will be required for efficient spectrum usage. Accurate channel state information will be critical to the correct allocation of resources.
Multiple and dynamic use of different modulation schemes. The diversity of use cases in 5G, well beyond those requiring high-speed data (4G), will be the driver of a wide range of modulation schemes.
Device-to-device communication facilitating network capacity off-load. Network architecture will have to be optimized to incorporate security, off-load potential, privacy, use of the network for acquisition and connection maintenance, and additional UE capability.
The many different use cases that 5G is accommodating suggests a set of network settings/functions, including radio access technologies (RAT) settings, for each use case—this is referred to as a “5G slice.” The idea behind the 5G slice is that the overall environment is maintained with only the necessary functionality for each use case to efficiently allocate available resources.
Given the technical challenges outlined above, we can draw some conclusions about end-user wireless devices, particularly mobile broadband phones. These devices will exist in an environment that includes:
- More complexity.
- More components, particularly MIMO and carrier aggregation (CA) filters.
- More demands on performance. High isolation between bands, low insertion loss (especially at band edges).
- Smaller size and lower cost. Overall phone size will not change significantly, nor will overall mobile device margin requirements. Thus, even though there will be more components built into the unit, they will have to be smaller and cheaper.
- Dual connectivity between cellular and Wi-Fi networks.
- Higher frequency components (>6 GHz) will be introduced as these components drop in size and cost.
A critical consequence of the increasing complexity and additional components is that the RF link itself will degrade, improvements in the overall performance of individual elements notwithstanding.
Key Technologies Needed to Make 5G a Reality
In order to meet, or at least approach, the challenging targets for 5G data rates, coverage, and connections, new technologies will be required. Identified technologies include:
High frequency spectrum (>6 GHz). With almost all the available spectrum below 6 GHz now allocated, carriers will be forced to move to higher frequency spectrum to secure bandwidth. However, as the frequency increases, RF propagation is reduced and penetration into buildings suffers. Thus, the bandwidths available above 6 GHz lend themselves to short-range, point-to-point, line-of-site connections (likely in-building).
Traditional network planning becomes difficult, but a much-higher-frequency, millimeter-wave (mmWave or near mmWave—e.g., 28 GHz) mesh network could work in dense urban environments. While components are generally smaller as frequency increases, reduced semiconductor performance at these higher frequencies will impact cost and power consumption.
Massive MIMO. MIMO is already used in LTE and LTE-Advanced networks, where multiple antennas are deployed either at the base station or mobile device (or both) to improve the link. However, massive MIMO is used in this context to leverage a large number of antennas to allow spatial beamforming, so that the energy can be focused directly at the user’s device and electronically steered to track with the device’s movements.
Large increases in capacity are achievable by generating multiple beam paths. Although this technology has been experimentally verified, significant advances in processing and interoperability will be required for wide-scale deployments.
Interference mitigation. 5G demands use of large spectrum bandwidths and massive overlapping of cells, which will result in both in-band and out-of-band interference. In general, 3G and early deployments of 4G were limited in their performance due to incomplete coverage. Performance within a cell was largely determined by signal-to-noise ratio (SNR).
The extreme densification of the network envisioned for 5G will change the limitations from coverage to interference, and performance will be determined by the signal-to-interference + noise ratio (SINR). RF filters will play a critical role in improving SINR. In 4G, cell edge performance in urban environments is dominated by co-channel interference from overlapping cells. Electrical antenna tilt can help, but inter-cell interference coordination (ICIC) has been introduced to reduce the impact on cell edge users.
ICIC uses different resource blocks for different users at the cell edge, which increases the effective reuse rate from 1 to 3. eICIC (enhanced ICIC) will be required for 5G, where resource block allocation will be dynamic in the time domain. This will be essential for heterogeneous networks, where small cells are within the macro footprint.
New and adaptable waveforms. OFDMA is the waveform used in 4G networks because, by using sub-bands, the bandwidth is easily scalable, enabling higher data rates. However, OFDMA is susceptible to interference and procedures to allocate resources are inefficient for small (data or packet) payloads. Extensive investigation of more efficient orthogonal as well as non-orthogonal waveforms is underway to identify candidates for 5G. RF filters will play a critical role in OFDMA performance.
Implications for the 5G RF Front-End
The current state-of-the-art for a mobile smartphone RF front-end separates the frequency spectrum into low-band (698 MHz-960 MHz), mid-band (1,710 MHz-2,200 MHz), and high-band (2,400 MHz-3,800 MHz) frequencies, which isolates the RF components, minimizes cross-talk, and optimizes the entire power amplifier-filter-switch path. Although integration of components is logical, the increasing complexity of 5G limits the number of manufacturers that have the expertise to execute on such a complex RF sub-system. Figure 1 shows a representative 4G smartphone front-end.
1. This is a current, state-of-the-art RF front-end architecture of a 4G smartphone showing filters, switches, and amplifiers.
5G RF front-ends for all wireless-enabled products will be driven by cost, power efficiency, and available space within the unit. They will need to be small, highly efficient, and able to be manufactured in large quantities to meet fast-growing global demand. To commercialize affordable custom parts for IoT devices in particular, RF front-ends will need to be designed with a minimum number of components, and manufacturing volumes will have to increase dramatically from current levels to reduce per-unit cost. In the current environment, most IoT devices are being built with low-cost parts originally developed for high-volume mobile phone production.
As we move toward 5G, the complexity of the RF front-end continues to increase. For instance, in addition to the main antenna path modules, diversity antennas provide both link robustness and increased downlink data rates. Designers are increasingly using receive diversity modules to process the diversity path, comprised of receive (Rx) filters and switches—and increasingly incorporating low-noise amplifiers (LNAs). Wireless carriers demanding higher 5G data rates will drive more carrier aggregation, creating more potential interference.
Consequently, the onus on RFFE designers moving forward will be to reduce complexity, reduce cost, while at the same time improving performance. These represent a significant set of challenges for RFFE design and manufacturing.
5G Filter Requirements
The growth in the sheer number of filters, and the ever more demanding filtering requirements, make RF filtering the critical pain point of the RF front-end. Unlike the power amplifier, where a single device can be used for multiple frequency bands and technologies, at the present time a single filter is required for each individual frequency band.
Although specific 5G requirements for filters will be developed in the future, the basic outline for those requirements has already become clear. It includes:
Complex multiplexing. Carrier aggregation will drive more complex multiplexing, which in turn drives more complex designs.
Increasing integration. Overall RFFE performance is crucial. Maximizing PA efficiency on the uplink, and receiver sensitivity on the downlink, will require optimization of the entire RF chain. As complexity increases, it will be crucial to understand the RF chain and any interactions between elements.
For example, consider the co-existence of band 5 and band 12, and the intermodulation distortion (IMD) that can interfere with other frequency bands (see table). Including IMD generation up to the seventh order, IMD products will land in five FDD receive bands, four TDD bands, 2.4-GHz Wi-Fi, and 5-GHz Wi-Fi. For filters, optimizing the interface to the PA and LNA will also be required in the filter design process.
Intermodulation products from band 5/band 12 and band 5/band 17. Green for FDD RX. Red for 2.4-GHz Wi-Fi/BT. Blue for TDD. Yellow for GPS. Orange for 5-GHz Wi-Fi.
Increasing number of filters. MIMO and spectrum proliferation will continue to grow and require more filters. Thus, the size and cost of filters must continue to decrease.
More demanding specifications. Isolation, loss, and power handling requirements continue to create new challenges. Filters in the RF chain are a major contributor to loss, which is critical for total transmit (Tx) efficiency (and ultimately for the current draw for the PA and battery life), and the total noise figure in the Rx path (and ultimately for the SNR and the data rate). Figure 2 shows a Tx configuration and related circuit attenuations.
2. Shown is a typical Tx path component line-up with estimated losses.
LTE, which is optimized for high-speed data, demanded significantly higher power than 3G protocols such as CDMA. And as such, the requirements for isolation and minimizing leakage into the Rx path (and vice-versa) grew. This will only be further exacerbated by high-power user equipment (HPUE), which uses more Tx power for improved cell edge coverage. In addition, power durability of progressively smaller filters becomes a major concern.
Frequencies greater than 6 GHz, and frequencies of operation greater than 20 GHz, will require different filter technology than the current acoustic wave filters used in mobile devices. Significant advances will be needed to reduce size and cost.
In combination, all of the above clearly shows that a 5G RFFE for mobile broadband will be extremely complex, and that the goal for filter design will be to both simplify the design process and the RFFE itself. Innovation that minimizes and reduces the complexity of components will be of paramount importance, and tools enabling this innovation will be critical to simplifying the RFFE.
Innovations that enable 5G RFFEs will need to include a low-loss triplexer (to minimize the number of antennas); multi-mode, multi-band PAs; and reconfigurable and multi-band filters (to reduce the number of filters and switches), all of which will need to be optimized as a complete system to reduce matching components.
Addressing 5G Design Issues
A new electronic system design platform is needed to develop the smaller, lower power devices that are needed for 5G designs. To date, each filter company has leverage in-house developed tools that are not designed to optimize filters for the constraints of 5G systems—nor do they have the accuracy needed to avoid costly and time-consuming board spins needed to perfect the design. The situation is analogous to developing circuit boards before the advent of today’s EDA tools. What's needed is a platform that can bring together:
- Modern filter theory.
- Finite element modeling (both electro-magnetic and acoustic).
- Novel optimization algorithms.
One example approach is that taken by Resonant Inc. The company has developed a tool called Infinite Synthesized Networks (ISN). ISN enables filter design teams to create novel filter structures that outperform traditional filter design. It occupies a smaller footprint and uses lower-cost technologies. In addition, ISN’s grounding in fundamental materials physics, while optimizing for high-volume design screening enables designs that are unconstrained by traditional acoustic wave filter design techniques. Figure 3 illustrates the design process with the tool.
Consequently, a designer using ISN can create multiplexers, wide passbands and high-power performance, and predict manufacturing yields as well, before a design is committed to mass production. Thousands of designs can be developed simultaneously and screened to maximize the ultimate performance of the device. Leveraging the expertise of filter design engineers for an increasing number of more complex designs can be achieved using ISN. This will be critical as the RFFE migrates to 5G in an environment where resources are constrained by a tight labor market of experienced design professionals.
4. Displayed are measured (blue trace) and modeled (green trace) duplexer performance.
ISN models are extremely accurate and reflect physical details of the filter structures, matching the measured filter performance not only in loss and isolation, but also in power handling and linearity. Figure 4 shows how closely ISN-modeled performance tracks the actual data measured on a Band 3 duplexer. The accurate modeling of acoustic filters using ISN enables the following:
Reduced development time and cost. Optimization is performed by the computer and the designer, rather than with expensive iterations in fab.
Solutions to challenging design problems. Empirical design techniques will ultimately be limited in headroom as more design parameters constrain the problem. This is particularly true for multiplexing and tunable filters. The flexibility and accuracy of the ISN tool suite is ideally suited for creating novel solutions to these increasingly complex requirements.
Design for multiple temperatures. The ISN framework allows optimization over multiple temperatures, optimizing performance at the higher temperatures that we can expect in a 5G RF module.
Designs optimized for high-power performance. LTE operates at higher power than CDMA, requiring designs that can withstand high power, at elevated temperatures, for extended periods of time. High bandwidth applications for 5G will continue this trend of higher power in order to achieve the highest data rates possible.
Improved yield and lower cost of production. ISN models help production engineers to create relevant fabrication parameters for reducing variability in the manufacturing process.
5G will drive up complexity in device designs, and that will require the industry to get serious about electronic system design tools that can help to develop smaller, more complex, and lower-power-consuming filters. The ISN is such a tool that promises to mix filter theory with substrate physics and algorithms to develop very accurate filters.
Dr. Bob Hammond is chief technology officer at Resonant.