The worldwide use of the radio spectrum is increasing at an astonishing rate. The demand for constant connectivity is driving the need for mobile broadband spectrum. Connectivity for typically underserved areas, such as rural areas and remote areas such as open ocean, is driving the development of extremely large constellations of broadband satellites, numbering in some cases in the hundreds of thousands. National defense drives the need for military radar, which often requires wide bandwidths to achieve its resolution and sensitivity objectives. Weather forecasting applications are of pertinent contemporary interest due to an anticipated increase in the number of severe storms, often attributed to climate change. These weather forecasting applications include weather radars as well as active and passive remote sensing systems (space-based and terrestrial). Broadband connections in homes and offices, including for applications such as Internet of Things (IoT), are often provided by unlicensed devices such as Wi-Fi, therefore driving the demand for more spectrum designated for unlicensed use. And the need for extreme bandwidths, such as for 5G systems (and beyond), is driving interest in traditionally underutilized portions of the radio spectrum, such as millimeter-wave (mmW, 30 − 300 GHz), and even terahertz frequencies (300 − 3,000 GHz).

The increase in spectrum use is producing an increasing number of potential conflicts among spectrum users. That is, one system operating on a particular channel in a particular band may be predicted to cause interference to another system that is operating on the same or other frequencies. The following types of interference may occur:

• Co-channel. Two systems operating on the same channel in a band.

• Adjacent-channel. Two systems operating on different channels in the same band.

• Adjacent-band. Two systems operating in immediately adjacent or relatively nearby bands.

• Spurious interference. Two systems operating in bands potentially far removed from one another, such as a victim system operating on a frequency that is a harmonic of the interfering system.

Incidences of concern over co-channel, adjacent-channel, and adjacent band interference are particularly on the rise as disparate systems are being required to share the same band, or operate in nearby bands, due to spectrum congestion and the lack of “clear” spectrum in which to deploy new systems or services. Therefore, an important component of improving the efficiency with which spectrum is used by current and future systems and services is to better understand co-existence among such systems and services. Here, co-existence refers to the ability of various systems and services to operate in the radio spectrum, within the confines of established regulations, without causing interference or burden on one another.

There are several regulatory constructs that relate to the concept of one system or service causing interference to another. It is useful to define these concepts before proceeding. The International Telecommunication Union (ITU) Radiocommunication Sector (ITU-R) and many national regulators use the following interference-related definitions [1], [2].

• “Interference. The effect of unwanted energy due to one or a combination of emissions, radiations, or inductions upon reception in a radiocommunication system, manifested by any performance degradation, misinterpretation, or loss of information which could be extracted in the absence of such unwanted energy.”

• “Harmful Interference. Interference which endangers the functioning of a radionavigation service or of other safety services or seriously degrades, obstructs, or repeatedly interrupts a radiocommunication service operating in accordance with [the ITU] Radio Regulations.”

• “Necessary Bandwidth. For a given class of emission, the width of the frequency band which is just sufficient to ensure the transmission of information at the rate and with the quality required under specified conditions.”

• “Occupied Bandwidth. The width of a frequency band such that, below the lower and above the upper frequency limits, the mean powers emitted are each equal to a specified percentage $\beta/2$ of the total mean power of a given emission. Note: Unless otherwise specified in an ITU-R Recommendation for the appropriate class of emission, the value of $\beta/2$ should be taken as 0.5%.”

• “Out-of-band Emission. Emission on a frequency or frequencies immediately outside the necessary bandwidth which results from the modulation process, but excluding spurious emissions.”

• “Spurious Emission. Emission on a frequency or frequencies which are outside the necessary bandwidth and the level of which may be reduced without affecting the corresponding transmission of information. Spurious emissions include harmonic emissions, parasitic emissions, intermodulation products and frequency conversion products, but exclude out-of-band emissions.”

• “Unwanted Emissions. Consist of spurious emissions and out-of-band emissions.”

Regulators often use the specific definition of “harmful interference” when determining whether one system or service is interfering (or predicted to interfere) with another. Unfortunately, the application of the definition of harmful interference is generally open for interpretation from an engineering perspective. Many terms that appear in the definition, such as “endangering,” “degrades,” “obstructs,” “repeatedly,” and “interrupts” are not themselves defined in any technical or mathematical sense.

Note that adjacent-channel and adjacent-band interference are typically caused by out-of-band emissions (OOBEs), while spurious interference is caused by spurious emissions. Co-channel interference is caused by signals whose necessary bandwidths overlap.

Regulators use various methods to mitigate the various types of potential interference. For example, the radiated power (Equivalent Isotropic Radiated Power, or EIRP) and occupied bandwidth are usually regulated to mitigate co-channel interference. OOBEs are almost always regulated to mitigate adjacent-channel and adjacent-band interference. Likewise, spurious emissions are usually regulated to limit harmonic emissions and other emissions in distant bands.

An important component of analyzing whether one system or service (the “interferer”) might interfere with another service (the “victim”) is the interference criteria that apply to the victim service. For example, a received signal that exceeds a level of $X$ dBm (or $X$ dBm/MHz) on the victim's operating channel, at the input to the victim's receiver, may be claimed to constitute harmful interference to the victim. That signal level may be specified as single-exposure (caused by one interferer) or aggregate (caused by all interferers combined).

Establishing reasonable interference criteria is important: criteria that are too lenient will not adequately protect the victim, while criteria that are too strict will place an undue burden on other spectrum users and result in underutilization of scarce spectrum resources.

A challenge with interference criteria is that they are typically established by the victim service itself, with little or no independent validation. For example, many services have established baseline criteria within the ITU process, in which work on service-specific matters is divided into numerous service-specific study groups and working parties [3]. Examples of interference criteria can be found in the ITU's collection of Recommendations [4]. While the groups invite inspection and comment from other groups, the reality is that each group is very busy with its own work, and often has relatively little time to validate the work of other groups. Further, many interference criteria that apply today were established decades ago, long before many spectrum conflicts existed or were foreseen, so various groups had less incentive to challenge interference criteria created by others, because they were not impacted by those criteria at the time. Nevertheless, established ITU-R interference criteria are adopted by many national regulators, without challenge or validation.

Lastly, interference criteria generally depend on the nature of the interfering signal. For example, the impact of a low duty cycle emission such as Wi-Fi (which typically transmits only a few percent of the time, but over a wide bandwidth) may be very different from the impact of an analog FM dispatch signal, which is a constant duty cycle signal confined to a narrower bandwidth. Interference criteria often fail to distinguish among the specific natures of the interfering signals.

Interference criteria are only one component of the coexistence analysis. Predicting how strong an interfering signal will be at the victim location relies upon one or more propagation models. Unfortunately, the state of propagation models is not particularly advanced, considering they are heavily utilized to establish spectrum policy and dictate spectrum utilization through regulations established to mitigate harmful interference.

As just one example, the Irregular Terrain Model (ITM) [5] is currently used in two modern sharing scenarios: the U.S. 3.5 GHz Citizens Broadband Radio Service (CBRS) [6], and the U.S. 6 GHz unlicensed band shared with fixed service links [7]. A particularly important component of ITM with regard to spectrum sharing is its prediction of long-path propagation loss caused by tropospheric scatter (troposcatter). ITM bases that component on the Longley-Rice troposcatter mode [8], which was developed based on empirical data acquired more than six decades ago (in the 1950s). Besides the data being acquired a long time ago, it is also sparse in frequency, geography, climate, and many other factors, calling into question its application today to modern and ubiquitous spectrum sharing applications across a wide range of frequencies.

The other challenge with ITM is that it does not include a clutter component, and close-in propagation essentially defaults to free space loss if there is no terrain blockage. In practice, this results in inappropriate predictions in heavily cluttered areas. For example, ITM sees downtown Manhattan as a flat piece of barren terrain, and has no knowledge of the myriad skyscrapers that block propagation paths. A comparison of ITM predictions to actual propagation in an urban environment shows that ITM under-predicted measured losses by an average of 29.8 dB, and that measured loss exceeded predicted loss more than 86% of the time [9].

Clearly, validation of propagation models, and establishment of new clutter-aware models, is a key area for development to improve the efficiency with which spectrum is used and shared.

The challenges and shortcomings of our incomplete understanding of co-existence among disparate systems and services has real-world consequences, as exhibited in the following examples.

All of the foregoing discussion points to the need for more robust analysis of the potential for harmful interference from one service into another. Typically, such analyses are conducted on an ad hoc basis, and only after conflict arises. Predictably, the interferer concludes that no harmful interference will occur, while the victim concludes precisely the opposite. Each sides' analyses are strongly informed by their respective agendas.

What is needed is the creation of one or more neutral third-party arbiters that are properly resourced to conduct the necessary technical studies such that they are able to adequately quantify the risk of interference on a statistical basis, from which informed decisions by regulators can be better made. The purpose of the work would be, in part, to provide mathematical and statistical meaning to the prediction of “harmful interference.” Such arbiters would fulfill the role of “co-existence umpires.” A coexistence lab would perform the following functions, among others:

• Hire skilled engineers and spectrum managers to design and conduct necessary studies

• Conduct laboratory tests to analyze and quantify the impact of various interfering signals to a victim system, considering co-channel, adjacent channel, adjacent band, and spurious impacts

• Conduct tests in environments consistent with actual deployments to determine how the laboratory results can best be applied to real-world scenarios

• Use the data to create a matrix of interference criteria that considers a variety of interferers and victims

• Work with experts in difficult-to-parameterize interference scenarios (for example, passive services) to analyze existing interference data to determine if such data can be used to empirically develop accurate interference criteria

• Conduct tests to validate propagation models, and help understand and quantify their limits. For example, conduct long-term tests of long-range propagation across a variety of frequencies, geographies, and environments, and examine the respective importance of troposcatter mode vs other longdistance modes, such as anomalous propagation by ducting, sporadic E, and other modes.

• Develop improved propagation models that take into account the improving quality and coverage of clutter data, such as building and foliage polygons being acquired by major tech companies.

• Conduct propagation tests to validate new clutter models.

• Publish the results of all studies

• Provide unbiased input to domestic and international regulators

Ideally, the work of a co-existence lab, as proposed, would be conducted by technical staff of regulatory agencies and other government labs. In reality, there are challenges with this approach. First, the government is typically under-resourced with appropriate technical staff. Such staff members are involved in numerous different proceedings at any given time, and there is not sufficient technical personnel to conduct a deep dive in all of the various proceedings simultaneously. Second, the regulator itself is also not always neutral. In some governments, the regulatory agencies are lead by political appointees who may have certain goals or objectives that they want to see implemented, and the staffs that report to such individuals are not always able to freely express their technical concerns about certain issues. Third, sometimes the need for co-existence analysis only arises at the time a regulator begins a proceeding on a certain issue, and the time to conduct thorough analyses may exceed the time that the regulatory agency has available to conclude the proceeding (for example, based on a legislative deadline). A co-existence lab will build a library of results that can be relied on when needed or, worst-case, the lab will be more nimble than the government and can conduct technical analyses more quickly than an under-resourced agency can.

The funding model and structure for a co-existence lab should be carefully considered. To remain neutral, a lab should not rely upon direct funding from industries with vested interest in the outcome of the lab's work. Ideally, the lab would be funded by the national government, either through the national regulator or through appropriate research funding agencies. Because the results of the work of a co-existence lab would be of international interest (due to the international aspect of spectrum use and regulation), perhaps a joint funding model among multiple governments, with the lab's work spread among multiple countries.

While the lab should not rely upon funding from entities with vested interest in its work, it should include an advisory council with representatives of all the different types of spectrum users. The lab would seek advice from such experts but would not be beholden to implement the advice of any specific interest.

The work of the lab should be open and transparent. Results should be published and freely available in the open literature. With appropriate constraints, observers should be allowed to view tests being conducted. The lab should be willing to consider collaborating with interested parties when appropriate, but the tests and their results should not be inappropriately influenced by any party, including national and international regulators.

While the lab should be thorough and accurate, it must be nimble, and it should avoid creating “analysis-paralysis.” It should also be willing to conduct revised studies whenever new developments arise, or whenever field experience may indicate shortcomings in prior results.

As spectrum becomes more congested and bands become more crowded, better analyses of the parameters under which different services and systems can co-exist are needed. Unfortunately, most such analyses today are conducted by the respective proponents. A respected third-party lab that can study co-existence issues from a neutral perspective is needed in order to help ensure that incumbent operators do not suffer harmful interference, while new entrants are not unnecessarily constrained such that scarce spectrum resources are wasted.