The wireless industry may be short of available spectrum below 6 GHz, but it has interference in abundance, and there’s even more coming in the future. It’s not just from 5G, the most likely culprit, but also from the enormous number of wireless-enabled devices employed by IoT networks, connected vehicles, and expansion of other services operating between about 500 MHz and 6 GHz. As always, RF and microwave filters and components formed from them will be effective tools for mitigating it.
To grasp the sheer size of the problem faced by wireless services and why band-pass and other filter types will be even more critical in the future, consider that even though 24 GHz frequencies to be used by 5G are just now being auctioned, issues are being raised about potential interference. The most recent come from the National Aeronautics and Space Administration (NASA), National Oceanic and Atmospheric Administration (NOAA), the Navy, and the American Meteorological Society (AMS). They believe that out-of-band emissions from 5G networks may interfere with the ability of satellite-based remote sensors operating at 23.8 GHz (Figure 1) to collect data about water vapor, which emits a faint signal that would be obscured by terrestrial signals.
The result, according to NOAA and NASA studies, would be partial to complete loss of data in some areas, degrading National Weather Service and Navy forecasts of tropical cyclones and other weather events. A similar problem was detailed in a study conducted by NIST in 2016 concerning interference from 4G signals to GPS receivers.
Tiny Signals, Big Challenges
The various short-range wireless protocols required for all IoT applications will often be used very close together, physically and electromagnetically, creating a classic interference scenario. Projections for how many “connected” devices will soon be in service vary widely, but are never less than tens of billions, and even these figures may ultimately prove conservative as connectivity is one of the major pillars of every IoT application.
Even though Bluetooth® and Wi-Fi have always had features that allow them to coexist in the same ISM band, they cannot cooperatively share channels. So, a Bluetooth device using one form of spread-spectrum modulation cannot detect and understand a Wi-Fi signal using another type when operating at the same frequency. The devices can be separated physically, in time, or in frequency, but when so many are operating near each other, no technology—even filters with very high rejection—can always ensure interference will not arise.
Collectively, the signals generated by IoT devices will make the airwaves even more congested, creating dense single environments concentrated in a few regions of the spectrum. Unfortunately, operators typically don’t know if, how much, and where interference will appear until the networks are deployed. Nor can they know whether harmonics and spurious signals from their own as well as other services will affect operation.
In a decade or perhaps a bit longer, another wireless application with enormous interference potential will appear, in the form of autonomous vehicles. To gain situational awareness, every vehicle will be equipped with at least one communications solution, and they all will communicate with each other, as well as with roadside infrastructure consisting of cameras and other sensors.
They’ll also need to connect with a network, which seems likely to be the solution provided by the cellular industry called “Cellular to Everything” or C-V2X (Figure 2), or possibly the Dedicated Short-Range Communications (DSRC) solution conceived in the 1990s and made an IEEE standard (802.11p) in 2009. In either case, operation will be in the 75 MHz of spectrum allocated for Intelligent Transportation Systems (ITS) around 5.9 GHz.
For vehicle autonomy to be fully realized, it must be deployed in all vehicles, or nearly so, and rely on millimeter-wave radar (as today), as well as C-V2X, GPS, and potentially other communications solutions, all but radar operating below 6 GHz. As spectral congestion in this frequency range is already very high, the addition of millions of connected cars, trucks, and buses will further increase interference. But in this case, interference isn’t just an annoyance but a criticality and cannot be tolerated.
Sharing a Finite Spectrum
As spectrum below 6 GHz is mostly spoken for, spectrum sharing is one of few alternatives for adding network capacity and the bandwidth required to accommodate very high data rates. Spectrum sharing isn’t new, but it hasn’t yet proven itself capable of reducing or eliminating interference between the shared services. With spectrum sharing, two or more services operate at the same frequencies and the incumbent service has absolute priority over the others and must be protected from interference.
When the second (or third) tier service detects activity on the same or possibly an adjacent channel, it must move elsewhere. In an ideal world, this would be an excellent solution for efficient frequency reuse, but the management systems required to achieve this are complex and in their most advanced form rely on transceivers located at every site that must be protected from lower-tier users.
A good example of spectrum sharing is the Citizen Broadband Radio Service (CBRS) created by the FCC in 2015, a three-tiered hierarchy offering 150 MHz of spectrum between 3550 and 3700 MHz. The priority or Tier 1 category consists of existing Navy coastal radars, some satellite earth stations, and a few other applications that cannot be affected by emissions from the other two– Priority Access Licensees (PAL) and General Authorized Access (GAA) users. The Tier 1 services are protected from interference by the lower two tiers, and the second tier (PAL) is protected from the lowest tier GAA users. GAA users have no protection at all. The entire network is managed by multiple Spectrum Access Systems (SAS) that enforce usage of the shared frequencies via cloud-based databases containing pertinent information about all CBRS devices.
The SAS has an enormous job. It must assign channels for CBRS devices and determine their maximum power at every location to ensure it is not exceeded. It also registers and authenticates devices, communicates with them for various purposes, resolves conflicting uses of the band, and receives and addresses reports of interference and requests for additional interference protection from Tier 1 users.
The sensors deployed at the Tier 1 sites detect interference and alert the SAS, which then commands the interferer to change channels, all in near real time. Even with the SAS hovering in the cloud, filters will be required in every end user device, base station, and Tier 1 transceiver, without which it would be nearly impossible to operate.
Carrier aggregation is one of the primary means of achieving greater bandwidth today and has increased from just a few channels in the downlink path to greater numbers in both downlink and uplink paths. Carrier aggregation has created several new challenges such as cross-isolation of aggregated frequencies and others. In the downlink path, the user device receives on two or more frequencies simultaneously, which creates the possibility of interference. The challenge is to ensure cross-isolation and in-band isolation on both the primary and diversity receive chains.
When channels are aggregated at frequencies far from each other, harmonic content at multiples of the fundamental (carrier) frequency can appear in the receiver, caused by nonlinear components in the RF chain. Solving this problem requires high levels of isolation between certain band combinations so that a harmonic of the lower-frequency band doesn’t degrade performance at the receive frequencies of the higher-frequency band.
Filters in the RF front-end are the solution to the problem and they must have very high rejection as well as low insertion loss, whether they are used in duplexers, diplexers, or bandpass filters. Carrier aggregation being used in the downlink path can interact between signals transmitted on other bands, creating intermodulation distortion in the receive bands of GPS (Figure 3).
Current 4G smartphones already support more than 30 frequency bands that require more than 60 filters. Consequently, miniaturization, multifunction integration, and reduction of components have become prime goals in the design of mobile devices, a task that has now been complicated by the addition of millimeter-wave frequencies. Millimeter-wave systems require low-loss narrowband filters that have extremely high selectivity and minimal temperature drift, and losses from the substrate are higher.
Considerable research efforts are being conducted to reduce filter size at millimeter-wave frequencies. Compound semiconductor technologies such as gallium arsenide and indium phosphide have better RF performance than CMOS, but they are more costly and do not allow the integration of digital, analog, and RF functions in a single IC. The latter is a major factor for smartphones as well as small cell base stations that use phased-array antennas with complete transceivers at each antenna element.
In the last decade, the interest in surface integrated waveguide (SIW) technology for realizing filters has increased dramatically (Figure 4), and it is likely that the technology will be a major contributor to millimeter-wave systems in the future. However, from a commercialization standpoint, SIW is still in its infancy and it will be years before integration of active components such as amplifiers, mixers, and AGC circuits will take years to develop.
SIW fits somewhere in between microstrip and dielectric-filled waveguide, in which a waveguide-type structure is fabricated and vias connect the upper and lower metal plates of a substrate. SIW combines some of the advantages of planar structures like microstrip and nonplanar structures like waveguide with guided-wave and mode characteristics like a conventional waveguide.
Losses with SIW are greater than classic waveguide because a dielectric other than air is within the guide, but they are still low. In addition, as the SIW structure supports only TM modes within the cavity, parasitic responses from unwanted TE modes are not present outside the passbands. When SIW achieves the ability to integrate passive components such as antenna elements and active components like amplifiers on a single substrate, the effect will be to eliminate transitions and their losses and parasitic effects.
SIW filters have even been fabricated on paper substrates and with additive manufacturing, for which the technology appears to be well suited. Using an inkjet printer, ink containing suspended silver particles realizes the top and bottom metal layers on the paper; the vias are made from brass rivets. SIW has also been fabricated using aluminum foils for the top and bottom sides of thick paper to define the top and bottom metallization of the SIW structure. Milling and the metal vias are fabricated using conductive paste.
Interference has been the bane of wireless communications since the earliest days of radio, and scientists and engineers of every era no doubt found their challenges to be the worst, and no doubt they were. But today, interference is making the design, deployment, and operation of wireless communications systems far more difficult than ever before. As new applications appear on already crowded bands and frequencies more than an order of magnitude higher are being explored, interference will become an even greater problem. Fortunately, the wireless community has thus far managed to keep interference mostly within acceptable levels, and RF and microwave filter technology is one of the foremost contributors to this success.