Better utilizing the resources available to us is one of technology’s greatest strengths. Whether it's getting more corn out of an acre or using our own white blood cells to fight cancer, technologists attempt to marshal the forces around us to achieve more with less. That’s especially true for resources that aren’t even tangible.
Case in point is the electromagnetic spectrum, which is used by everything from implantable RFID chips in pets to flying cars. However, it is almost criminally underutilized because of outdated technology and regulations. That’s where cognitive radio technologies come in. They represent an effort to fully utilize the electromagnetic spectrum, and in doing so dramatically increase the speed of everything from traffic systems to health imaging.
Crowded airwaves
The electromagnetic spectrum spans the entirety of frequencies with which electrons can move. However, some ranges are more valuable than others due to the ease of creating systems that can create, modify, and receive signals in them. That makes those bandwidths a limited resource with plenty of potential demand for everything from amateur radio users to military communications lines.
A limited resource and multiple demands on it begs for regulation, so in most geographical locations a governmental or quasi-governmental agency has stepped in to regulate the spectrum by statically defining what bandwidths can be utilized by what kind of users. For example, in the U.S., amateur radio sets can operate in any of several bandwidths, such as 135.7 Hz to 137.8 Hz, but are forbidden from operating outside of those narrow bands.
This creates a situation where some bands, such as those utilized by cell phones, are jam packed and have almost unlimited demand and are therefore bandwidth limited, while others, such as those used for emergency responders, are typically empty but may need to be utilized in a literal life or death situation.
A potential solution would be to develop a protocol that understands the different requirements for different bands, and utilizes an adaptable, intelligent framework to fully utilize all of the bandwidth available in the spectrum while keeping it clear for necessary high priority users and eliminating crosstalk noise.
Building the intelligent radio
Enter cognitive radio (CR) — a technology built on top of another, underlying technology known as software defined radio (SDR), which itself has been gaining traction in wireless communication research. The idea for SDR was originally introduced in 1992 by Dr. Joe Mitola III. The basic function of SDR is to modify the parameters of a radio network by modifying software rather than the physical components of the radio itself, such as antennas and frequency shifters.
As SDR became more widely adopted, Dr. Mitola added another layer — CR. SDR at the time was statically defined by the algorithms it was programmed with, whereas CR would utilize adaptive technologies like artificial intelligence and machine learning to modify its own parameters based on the operating environment it found itself in. Ideally, this would allow it to fully utilize the spectrum available to it in its specific geographical location, while also interfacing with other CR systems to ensure the highest possible bandwidth transmission while eliminating noise.
Underpinning CR are two separate techniques that both rely on the fundamentals of SDR to function. First, the system must be able to detect unused frequencies in its local area, known as “spectrum sensing.” Then, the CR system must be able to adaptively modify its frequency and power output depending on the changing situation it finds itself in, known as dynamic ppectrum access (DSA).
Spectrum sensing involves monitoring the local electromagnetic environment and looking for “white spaces” where there are minimal or no other users. These blank spaces in the spectrum represent potential underutilized spectrum, and therefore, they are a potential resource the CR system can exploit.
Researchers have developed a wide variety of techniques to perform this task, ranging from simple energy monitoring (lower energy = fewer users) to cyclostationary-feature detection which looks for unique patterns present from commonly used wireless protocols like quadrature phase-shift keying (BPSK) to determine how many signals are being transmitted on the line. Unfortunately, there are plenty of challenges facing any algorithm that attempts this task, and research is ongoing to optimize those algorithms.
DSA is somewhat simpler from a technical standpoint, as it involves a series of logical checks to determine an optimal match of frequency and transmit power — essentially an optimization problem based on the input from the system’s spectrum sensing algorithms. But it also must be capable of continually finding that optimal solution, as it might change over time depending on the dynamic situation in the area. If an emergency responder suddenly joins a bandwidth, any CR operating there must abandon that frequency or at the very least limit its output power.
The underlying technology underpinning these two features is AI, or, back when Dr. Mitola originally developed the idea, machine learning (ML). Adaptive learning, which is such a critical feature of CR, is something that AI excels at, and, therefore, as AI improves, so does CR. Pattern recognition in spectrum analysis or continually evolving the optimization algorithm for DSA are bread and butter to an AI, even if they’re not quite capable of offering the optimal (or correct) solution all the time yet.
Where cognitive radio shines
Wireless communications networks are always evolving, with the next step being “beyond 5G” or “6G” depending on the source. These next generation technologies utilize bandwidth more intensively than past generations, including moving data transfer into frequencies never before attempted, such as terahertz and millimeter waves.
But perhaps the most striking use case for CR is the proliferation of sensors in almost everything. The Internet of Things already consists of billions of connected devices, all attempting to talk over the same frequencies, and in many cases in the same spatial area. Locations like manufacturing plants, data centers, and warehouses can contain hundreds of thousands of sensors all vying for bandwidth, and all confined to a small area. CR’s capabilities in spreading the data across a wider spectrum, and ensuring that the devices don’t interfere with one another, will be critical to the long-term adoption of technologies that require higher data throughput for myriad sensors in constrained spaces.
Another obvious use case for CR is the military. Adaptive networks, such as the mobile ad-hoc networks (MANETs), already account for the variability that wireless networks currently experience in battlefield environments. Drones, operators, sensors and weapons are all moving around a dynamic space, with new information flooding in from a wide variety of sources. In order to make sense of it all, military commanders have to get the important data fast, and then react to it.
CR can help ensure that, no matter the physical implementation of the network, all the data available can make it back to the command and control center, and vice versa by allowing stable connections from commanders to go back out to mobile platforms like UAVs or field units, without data loss or interference.
There are plenty of other use cases for the CR, but one that is more out-of-this-world is its use in space applications. NASA has been looking at CR to solve some of its problems in communicating with deep space probes, which includes the fact that the Earth both rotates once a day, and also rotates around the Sun, creating various interference and disruption patterns in its communications links with those craft.
The agency has developed platforms like the space communications and navigation (SCaN) testbed on the International Space Station to validate some CR technologies. Other space users have focused on the integrated satellite-terrestrial communications network, which is a CR enabled system that increases the bandwidth utilized by satellite downlinks and uplinks, which will become increasingly critical as more and more satellites are launched by ever-cheaper rockets.
A cognitive radio concept architecture that makes decisions based on various inputs and a flexible SDR unit whose operating software provides a range of possible operating modes. Source: Wireless Innovation Forum
Navigating the hurdles
Plenty of barriers still lay ahead for the full adoption of CR, some more tractable than others. The technology’s unique spectrum-hopping capabilities make it vulnerable to specific kinds of attack that would affect other kinds of radios. However, perhaps the biggest obstacle to CR adoption is self-imposed — regulatory bodies still explicitly control the spectrum in much of the world, and without a change in the rulesets, there will always be limits to how much CR can actually do to improve bandwidth.
If a bad actor knows a system is using CR, they can implement a series of attacks that can limit the effectiveness of the system, though not necessarily block it entirely. These attacks can range from primary user emulation (PUE), where a bad actor pretends to be a primary user of a band (like the emergency responder’s band), thereby blocking CR’s access to it, to spectrum sensing data falsification (SSDF) where a malicious actor attempts to send incorrect spectrum data back to the CR sensing, thereby confusing the optimization algorithm and limiting its effectiveness.
However, the real headache for CR development is the slow pace of regulatory change. Many spectrum controls are based on databases that are only updated when a regulatory body, such as America’s Federal Communications Commissions (FCC) or the International Telecommunications Union (ITU) decides to change them. Such static definitions are anathema to an adaptive system like CR, so until the regulatory bodies see the potential of the new technology or are convinced that there’s a problem they need to solve with bandwidth, the flexibility that would be required to make CR viable commercially will not be available.
Unlocking the spectrum
Even if there will be some waiting involved, something akin to modern CR will eventually be widely adopted. The advantages far outweigh the costs — CR promises increased bandwidth and lower interference in a more interconnected world. While there are risks, some of them can be solved by even more technological advancement, such as quantum computing systems or blockchain technologies.
As all these technologies become integrated together, the world of wireless communications begins to expand — and CR will play a central role in that process, even if the regulators have to be convinced of its value first.
About the author
Andy Tomaswick is an engineer and freelance writer who’s passionate about education, space exploration and making the world better through technology. When not engineering or writing something, he spends time with his family or running in circles to stay in shape.
