Mind Matters Natural and Artificial Intelligence News and Analysis
vintage-microphone-and-headphones-with-signboard-on-air-broadcasting-radio-station-concept-stockpack-adobe-stock.jpg
Vintage microphone  and headphones with signboard on air. Broadcasting radio station concept.

We Are Running Out Of Radio Spectrum But AI Might Help

Future generations of wireless (radio wave) communications will depend on artificial intelligence (AI)

Today’s wireless (radio wave) communications are already vastly different from those of previous generations. One difference is the move to much more adaptable hardware controlled by software—software defined radio or reconfigurable radio. Future generations of the technology will move increasingly to dynamic spectrum networks (DySPAN), using adaptive spectrum access techniques.

One outcome might be an improvement in transmission priority. Most people would agree that a call to 9-11 should get priority routing as opposed to routine calls to friends and businesses. The networks will reason not only about its transmission options but the needs of the underlying content. AI enables these possibilities.

But here’s the problem: Based on traditional methods of spectrum management we have run out of spectrum.

The traditional method for separating different radio services, often called “command and control” spectrum management, separates services based on frequency and distance. Two radio stations can use the same frequency if they are far enough apart. Radio stations that are close to each other will be assigned different frequencies.

As all available frequencies were being used, the FCC formed a Spectrum Policy Task Force, with the mission of finding a solution to this problem. The conclusion of the FCC’s task force was that the fundamental problem was not a scarcity of spectrum but the method being used to manage it. Spectral studies found that most frequencies went unused most of the time. Almost all frequencies were reserved for one or more radio services. Others could not use them, even if the services they were reserved for were not using them. From these studies came the concept of dynamic frequency allocation. Why not let others use a frequency, if the service it was reserved for was not using it at a particular place and time?

In order to change to this new dynamic method of spectrum management, radio service providers would need to monitor the frequencies they plan to use and coordinate their use of a frequency channel with other users. Fortunately, radio has developed from primarily hardware devices to software-controlled devices. Improvements in radio frequency (RF) components make it possible to design highly adaptable RF radio sections that are defined and limited by the software that controls them. This change to software-controlled or software-defined radio enables a device to adapt to its spectral environment. In these new devices, data from spectrum sensors is analyzed and a transmission frequency selected. Only after a device has determined that it will not interfere with other users of the spectrum will it transmit a message.

This relatively simple conceptual change comes with impressive challenges. If new devices can change the frequency they use, how can a regulator certify it, knowing that it will not transmit in a restricted band? How will a device near the border know whether to use the US or Canadian regulations to govern its transmission? How will sensitive radio services know they will not be interfered with by these new highly adaptive devices? The regulator’s job has been made much more complex.

Thus, one outcome of the move from ”command and control” to dynamic spectrum management is the introduction of risk management and the use of ontologies. In the past, with hardware-based devices and “command and control” spectrum management, there was risk. The system was not perfect and interference did occur. The change to dynamic spectrum access forces a new look at the risks being introduced and a difference approach to managing those risks. That change and other factors draw in ontological considerations.

In metaphysics, ontology is the study of the nature of being. In its more specific application to spectrum management, it is the study of the objects, concepts, and categories in a domain and the relationship between them. As we will see, ontologies are usefully applied on many levels. For spectrum management the ontology of interest includes all users of spectrum, intentional and unintentional transmitters, but also receivers. The first object of study is how different wireless devices might share the same spectrum on a non-interfering basis. However, the study must go much wider and include the potential impact of transmissions on unintended receivers.

A regulator must understand the new set of risks and the interaction of these factors and their boundaries. New regulatory methods address the new situation and be implemented in ways that adequately safeguard the public interest.

This safeguarding of the public interest represents a moral value. The decisions made are no longer technical, although technology informs the implementation of the decision. A pattern will be found in which the final decisions repeatedly are value judgments based on the decisionmaker’s morals or values. This in turn raises the question of where those morals or values come from and when there are different value systems, what decides which one will guide decision making?

The initial uses of dynamic spectrum access use relatively simple concepts. There may be rules that a device must listen to the frequency it intends to use and make sure it is clear before it can use it, listen-before-talk. From a simple base like listen-before-talk more complex methods can be developed. A next step might be to require systems to monitor a number of frequency options and use the one most distant from other users, least-interfered-channel.

Another common approach is to have an authoritative database of protected locations. Before a device can transmit it must check its location against the database of protected locations. It can only transmit after it confirms that it is not near one of the protected locations.

Many more complex rule sets can be developed to optimize the complex use of spectrum of many users and services sharing the same spectrum. This approach to spectrum management is called policy-based radio because it manages the spectrum based on a set of policies.

Conclusions

New methods of spectrum management are being implemented because the traditional methods have proven to be inefficient and wasteful of spectrum. The move to policy-based radio is enabled by improvements in RF components, processors, and software. Devices that in the past were simple and static and now highly adaptable and being required to make complex decisions. Advances in AI enables even greater sophistication in the future. New technologies enable very sophisticate decision making advancing the values and priorities controlling those decisions. The question becomes, what values and priorities will be used to develop the future of wireless communications?


Stephen Berger

Stephen Berger is the founder of TEM Consulting, LP. He specializes in developing consensus multidisciplinary solutions for complex public policy issues. He has served on three federal advisory committees. Two of those committees addressed accessibility of telecommunications and information technology for people with disabilities. The third addressed requirements for voting equipment. He has chaired five standards committees that developed standards incorporated by the FCC and FDA into the US Code of Federal Regulations (CFR). A current focus of this work involves improving healthcare through insightful introduction of technology with supporting system change.

We Are Running Out Of Radio Spectrum But AI Might Help