by Barry Manz, Editor, Microwave Product Digest
Since the dawn of radio and communications, interference has been an enormous problem that requires multiple measures to solve. As the airwaves become more densely packed with signals using higher-order modulation schemes intolerant of interference (and much else), it’s becoming even worse. Defense systems from EW to radar, communications, and navigation face unique challenges that their commercial wireless and broadcast counterparts do not. The warfighter is thus presented with a wall of electromagnetic energy from which it must attempt to separate a desired signal or signals from the rest.
Interference comes not just from adversaries but friendly forces, from equipment interfering with itself, and from passive intermodulation distortion (PIM) that is often not only exceedingly difficult to locate but appears seemingly out of nowhere. Facing challenges like this—and more—it is no wonder that DoD is addressing interference mitigation more aggressively than ever.
At a high level, the problem appears simple: keeping interfering signals from degrading receiver performance, but achieving that is extremely difficult. For example, transmitters cause intermodulation distortion, harmonics, and spurious emissions that overwhelm the receiver’s front end. When this occurs, its signal-noise ratio is degraded that in the worst cases makes it impossible for the receiver to detect the desired signals.
Well-known techniques for reducing these emissions include operating amplifiers below their saturation point (back-off), as well as analog and digital pre-distortion and reducing phase noise. Increasing the linearity of the receiver also aids in preventing intermodulation products from interfering with the desired signals.
In the receive chain, the most effective solution for reducing interference has long been the venerable analog bandpass filter, whose ability to reject out-of-band signals is exceptional. The bandpass filter (Figure 1) passes frequencies within a narrow range while rejecting (attenuating) frequencies outside that range. The level of rejection can be very high, 50 dB or more, effectively reducing the signal strength of interferers to manageable levels. But a bandpass filter rejects signals only over a specific range of frequencies, so switched filter banks are often used to cover multiple frequencies, and they can become large. In contrast, digitally tunable filters provide greater protection over wider bandwidths.
Considerable research is being conducted into filters today and in addition to interference cancellation are the primary mechanisms by which the most advanced interference mitigation systems are constructed. Remedies such as these can be applied only to systems over which the owner has control and are not applicable to those of adversaries or even operators of licensed and unlicensed systems. As a result, they represent a solution to just one set of interference problems but are more than enough to keep designers up at night.
There is also the onerous problem of interference caused by PIM that is different from other forms of intermodulation distortion. It is not generated by active (non-linear) components like amplifiers but by components normally considered linear, such as antennas, attenuators, cables, duplexers, diplexers, filters, and connectors, in which oxidation or other effects may cause them to be non-linear.
PIM appears when two or more signals are present in a passive non-linear device that mix and produce other signals related to them. Unlike mixers, in which such a result is desired, the result in this case is undesirable. If these signals are significantly weaker than the desired signal they may have no effect on a receiver, but if they occur after combining, for example, they can reach levels high enough to form distortion products at amplitudes high enough to degrade receiver performance. While active components like amplifiers can produce extremely high levels of distortion, it can be removed by filtering, while PIM typically cannot because they can be generated late in the transmit signal path or outside the system entirely.
PIM has typically been considered a problem for systems generating high RF power levels, a misnomer that has remained in its definition primarily because high-power systems, such as Navy ships and AM, FM, and TV broadcast installations, were the first to experience it. In practice, studies have shown that components producing PIM can cause damaging effects several miles away, even when their signal strength is low.
The problem on Navy ships has been so severe for so long that it received its own moniker—the Rusty Bolt Effect (Figure 2), in reference to the corrosion and rust present almost anywhere on a ship where interactions of electromagnetic energy with dirty connections or corroded parts cause the same effect as a diode. As more and more discrete transmit and receive systems have been added “topside” over the years, in combination with the ship’s structure, they are a breeding ground for PIM generation.
The wide variety of ways PIM can be generated makes it extremely difficult to detect, but over the last 15 years or so, portable PIM analyzers have made this much easier. Resolution of the problem typically requires inspection and repair or replacement of the offending component, or when generated by a ship’s infrastructure, can require extensive measures such as repainting large surfaces of the superstructure.
Generally speaking, there are two primary means of achieving interference mitigation: filtering and interference cancellation alone and together. The solutions described later in this article use either one or both. However, as noted earlier, the design of receivers is changing rapidly from their traditional architecture that relies on analog components, including a mixer and local oscillator in the front end, to direct RF sampling that eliminates them in favor of digitizing the incoming signal directly when it is captured over the air. While this approach has enormous benefits, it also introduces a problem related to the use of very high-performance analog-to-digital converters.
Direct RF sampling has become the architecture of choice for applications ranging from EW to SIGINT, COMINT, ELINT, radar, and communications. As the approach eliminates nearly all RF components in the signal chain before the input signal is digitized, a system can be smaller and less complex because once in the digital domain, functions traditionally performed by large analog components can be performed digitally in a digital signal processor or an FPGA.
In these receivers, high performance can only be obtained when its signal-to-noise ratio (SNR) and spurious-free dynamic range (SFDR) are extremely high. The component most important for achieving this is the ADC because, as the first signal processing component after signal capture, it defines the performance that the entire receiver can achieve.
One of the most common and useful measurements of ADC performance is its effective number of bits (ENOB). Even the most impressive of today’s devices pose significant design challenges, especially as sampling rates and instantaneous bandwidth increase and when time-interleaving of ADCs is employed. Unfortunately, an ADC has errors in quantization, offset, gain, linearity, and timing that create spurious signals in its output.
If the strength of these signals is high enough, it becomes difficult and sometimes impossible to separate the signals of interest from the noise. Techniques such as clock dithering, calibration, and commutating the ADC at lower rates have been used to mitigate the issues in the ADC, but each one has significant drawbacks that often cause as many problems as they attempt to solve, and require considerable computing power, as well.
One of the most recent approaches to solving this is the High Dynamic Range Receiver (HDRR) solution developed by Precision Receivers. HDRR does not have the shortcomings of other methods, according to the company, and is most effective when acting on signals with high dynamic range and wide bandwidth. It can achieve an order-of-magnitude improvement in reducing unwanted spurious signals, improving spurious-free dynamic range (SFDR) by up to 15 dB and optimizes the ADC’s ENOB as well.
HDRR can be used in any direct-sampling system regardless of its ADC and at any frequency of interest. It also does not require self-calibration, reduces anti-aliasing filter complexity, and minimizes the required amount of post-processing and signal analysis. It is based on the fact that an RF input signal from an antenna consists of desired signals and noise and that digitization in the ADC introduces another noise-distortion component.
HDRR modifies this additional noise component by manipulating the ADC’s control signals using an approach developed by the company over several years. The process effectively removes the noise contributed by the ADC, producing an output signal with much less distortion, which is then passed to an FPGA for processing or to a mass storage device, depending on the application (Figure 3).
Another Problem to Solve
Defense and commercial wireless systems face the challenge of increasing spectral efficiency, as there is precious little spectrum to waste within the communication sweet spot between about 600 MHz to 7 GHz. The “ideal” of achieving greater spectral efficiency is full-duplex rather than half-duplex communication. This allows transmission and reception to be performed simultaneously in the same channel, effectively doubling spectral efficiency as the bandwidth can be used simultaneously for both transmission and reception.
However, achieving full-duplex operation is extremely challenging. Due to the emergence of 5G and the requirements of next-generation defense communications systems, it has become an extremely important problem to solve. The result has been a flurry of journal articles extolling the virtues of various approaches taken by the wireless industry and academia.
The most difficult challenge is the self-interference from the transmitted signal power to the simultaneously received signal. Unfortunately, this creates an overlapping of the strong signal with the much weaker received signal of interest, producing considerable self-interference. This signal can theoretically be removed by subtracting the transmitted signal from the received waveform, but the signal will be linearly and nonlinearly distorted while propagating to the receiver.
The problem is the result of RF power amplifier non-linearities, transmitter and receiver in-phase/quadrature (I/Q) imbalance, the phase noise of the local oscillator, and ADC quantization noise. To be effective, an approach must ensure that the self-interference power level is below the noise floor by at least 70 dB.
If the approach works, the system can cancel its own transmitted signal in its receiver, and what it transmits does not impact what it simultaneously receives, which requires that changes in the time-varying self-interference channel must be tracked in real-time, so the system must be self-adaptive. Some of the most impressive work has been conducted at Stanford University, whose researchers have achieved self-interference cancellation greater than 100 dB and its architecture is widely used as the foundation of other approaches.
Full-duplex capability will have a profound effect on spectral efficiency, and it could not come at a better time for both the wireless industry and DoD that are inextricably intertwined. DoD “owns” an enormous amount of spectrum, some of which it has reluctantly offered to the FCC for auction, which will help the wireless industry in the U.S., where there is less unused “mid-band” spectrum available than in almost every other first-world country. Every bit of it will be required for 5G to be realized.
These efforts toward interference cancellation are far from the only ones either in progress or in service. For example, the goal of DARPA’s Wideband Adaptive RF Protection (WARP) program is to “harden” wideband receivers operating in congested and contested EM environments using adaptive filters and signal cancellers that selectively attenuate or cancel signals (Figure 4). In addition to intention jamming, it addresses interference caused by a transceiver’s transmitter, sniffing the EM environment and using wideband tunable filters to maintain the receiver’s dynamic range without decreasing its sensitivity. That is, a WARP solution would reduce the effect of strong signals on the desired signal without attenuating without reducing the receiver’s performance.
In addition to filtering, the program also addresses adaptive signal cancellation technology. The traditional means of doing this is by using a duplexer to separate signals on different bands but is not effective when radios transmit and receive on the same frequency to increase spectral efficiency. Unfortunately, this approach, called same-frequency simultaneous transmit and receive (STAR), has the risk of self-interference caused by the transmitted signal unintentionally disrupting the receiver input.
Conventional techniques for self-interference mitigation in these situations require minimizing the coupling between the transmitter and receiver or employing a controllable auxiliary path between the two ports to cancel the signal coupling. Still, these techniques sometimes fail to achieve the required amount of cancellation and are not reconfigurable or scalable across a wide range of frequencies.
Researchers have already proposed using a full-duplex antenna that achieves self-interference cancellation through polarization duplexing, and the WARP program is attempting to improve on this by sampling and cancelling it in the digital domain. The process is already being utilized by several companies, such as L3 Harris, which has been working on interference mitigation for many years. Its solution employs various tools depending on the type of threat, such as multiplexing and RF filters and interference cancellation.
The latter samples interfering signals and creates the opposite of it in real-time, and when combined they cancel each other out, leaving only the desired signal. It also cancels out noise and spurious signals even if they are on the same frequency. The result can be 100 dB of interference cancellation. Its latest versions use the company’s Advanced Interference Mitigation System (AIMS) that makes it possible to quickly enable solutions for a wide range of interference profiles.
Cobham’s mINCAN interference cancellation system injects an anti-phase version of the interfering signal, like the approach used for acoustic noise cancellation in smartphones, headsets, and hearing aids. However, this system can handle very high levels of received interference and accommodates the high speeds of frequency-agile transmitters.
A sample of the signal from the transmitter in a co-located situation is taken using a directional coupler in the transmit antenna path and is scaled in phase and amplitude before being mathematically added to the receive path to cancel out the interference. Multiple interferers can be cancelled by allocating a module to each one, and cancellation occurs before the interfering signal reaches its full amplitude to avoid receiver blocking.
Another technique developed at MagiQ called Agile Interference Mitigation System (AIMS) identifies high-level signals that would otherwise lead to distortion in the receiver and sets up frequency-agile, very-high-Q filters to quickly suppress them. The AIMS device is located between the antenna and receiver and discriminates signals of interest from interferers and suppresses the latter signals before reaching the receiver without effect on the desired signal.
AIMS monitors the spectrum in real-time, identifies unwanted interference signals and invokes a very high Q filter to suppress the problem frequencies. These filters allow removal of about 60 dB of interference while leaving nearby signals of interest uncompromised. The tunable filters can be continually updated to adapt to rapidly changing EM environments.
Septentrio’s AIM+ system is focused on satellite-based navigation systems and mitigates the effects of narrowband interference using three notch filters that remove a narrow part of the RF spectrum around the interfering signal. The power spectrum plot created by AIM+ technology detects and neutralizes interference, resulting in faster set-up, reduced downtime, and secure operation. AIM+ protects against simple narrowband interference as well as more complex wideband interference, including jamming and spoofing.
Kumu Network’s self-interference cancellation technology is also designed to allow radios to transmit and receive on the same or adjacent channels. Its canceller suppresses the interference a transmitter presents to a co-located receiver even if the two radios operate with no guard band between them. The solution adapts in real-time to the changing environment to ensure consistently high isolation between the two radios.
The technology is also available in chip form with frequency-agnostic taps for self-interference cancellation or FIR filtering. Digital cancellation taps can be used where analog cancellation alone does not provide sufficient cancellation or where longer delay reflections impact the receiver must be canceled. It has four FIR filter chains organized as 12 programmable taps per chain with a maximum of 350-ns aggregate delay through each chain.
Two chains can be cascaded for a maximum of 24 taps for a total delay of up to 700 ns, or the chains can be configured to support 2×2 MIMO operation. The IC supports a range of signal processing applications where analog signal manipulation is required to avoid the delay and resolution problems that digital conversion introduces. For example, the chip can be used to implement full-duplex systems or to improve co-existence between co-located radios.
At some time in the very (very) distant future, all types of interference may somehow be resolved, but until then, DoD, industry, and academia will continue to throw every possible resource into reducing it. It will not be easy, as there are so many types available to work on at every level.
Those interference cancellation techniques described in this article may be the most widely used or potentially promising, but there are others as well, and with so much work being conducted to eliminate it, new ones will continue to be continuously revealed.