PreprintPDF Available

A Semantic Framework for Enabling Radio Spectrum Policy Management and Evaluation

Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Because radio spectrum is a finite resource, its usage and sharing is regulated by government agencies. These agencies define policies to manage spectrum allocation and assignment across multiple organizations, systems, and devices. With more portions of the radio spectrum being licensed for commercial use, the importance of providing an increased level of automation when evaluating such policies becomes crucial for the efficiency and efficacy of spectrum management. We introduce our Dynamic Spectrum Access Policy Framework for supporting the United States government's mission to enable both federal and non-federal entities to compatibly utilize available spectrum. The DSA Policy Framework acts as a machine-readable policy repository providing policy management features and spectrum access request evaluation. The framework utilizes a novel policy representation using OWL and PROV-O along with a domain-specific reasoning implementation that mixes GeoSPARQL, OWL reasoning, and knowledge graph traversal to evaluate incoming spectrum access requests and explain how applicable policies were used. The framework is currently being used to support live, over-the-air field exercises involving a diverse set of federal and commercial radios, as a component of a prototype spectrum management system.
Content may be subject to copyright.
A Semantic Framework for Enabling Radio
Spectrum Policy Management and Evaluation?
Henrique Santos1[0000000221106416], Alice Mulvehill1,3 [0000000329810675],
John S. Erickson1,2[0000000330784566], James P.
McCusker1[0000000310856059], Minor Gordon1 [0000000319284130], Owen
Xie1[0000000286893385] , Samuel Stouffer1[0000000289295989], Gerard
Capraro4[0000000218369231] , Alex Pidwerbetsky5[0000000174368105] , John
Burgess5[0000000195396922] , Allan Berlinsky5[0000000153147178] , Kurt
Turck6[0000000192240164], Jonathan Ashdown6 [0000000172021095], and
Deborah L. McGuinness1[0000000170374567]
1Tetherless World Constellation, Rensselaer Polytechnic Institute, Troy NY, USA
2The Rensselaer Institute for Data Exploration and Applications, Troy NY, USA
3Memory Based Research LLC, Pittsburgh PA, USA
4Capraro Technologies Inc., Utica NY, USA
5LGS Labs, CACI International Inc., Florham Park NJ, USA
6Air Force Research Laboratory, Rome NY, USA
Abstract. Because radio spectrum is a finite resource, its usage and
sharing is regulated by government agencies. These agencies define poli-
cies to manage spectrum allocation and assignment across multiple or-
ganizations, systems, and devices. With more portions of the radio spec-
trum being licensed for commercial use, the importance of providing
an increased level of automation when evaluating such policies becomes
crucial for the efficiency and efficacy of spectrum management. We in-
troduce our Dynamic Spectrum Access Policy Framework for support-
ing the United States government’s mission to enable both federal and
non-federal entities to compatibly utilize available spectrum. The DSA
Policy Framework acts as a machine-readable policy repository provid-
ing policy management features and spectrum access request evaluation.
The framework utilizes a novel policy representation using OWL and
PROV-O along with a domain-specific reasoning implementation that
mixes GeoSPARQL, OWL reasoning, and knowledge graph traversal to
evaluate incoming spectrum access requests and explain how applicable
policies were used. The framework is currently being used to support live,
over-the-air field exercises involving a diverse set of federal and commer-
cial radios, as a component of a prototype spectrum management system.
Keywords: Dynamic spectrum access ·Policies ·Reasoning
?Approved for public release (reference number: 88ABW-2020-1535).
arXiv:2011.04085v1 [cs.AI] 8 Nov 2020
2 H. Santos et al.
1 Introduction
Usable radio spectrum is becoming crowded7as an increasing number of ser-
vices, both commercial and governmental, rely on wireless communications to
operate. Techniques known as Dynamic Spectrum Access (DSA) [26] have been
extensively researched as a way of promoting more efficient methods for sharing
the radio spectrum among distinct organizations and their respective devices,
while adhering to regulations.
In the United States, spectrum is managed by agencies that include the Na-
tional Telecommunications and Information Administration8(NTIA) and the
Federal Communications Commission9(FCC). The NTIA publishes revised ver-
sions of its Manual of Regulations and Procedures for Federal Radio Frequency
Management10 (commonly referred to as the NTIA Redbook) which is a com-
pilation of regulatory policies that define the conditions that non-US, as well as
federal and non-federal US, organizations, systems, and devices must satisfy in
order to compatibly share radio spectrum while minimizing interference.
With the advent of 5G,11 more parts of the radio spectrum are being licensed
for commercial usage and, with the increased availability of cognitive radios [25]
(devices that are able to automatically adjust their operating frequency), the im-
portance of providing an increased level of automation when evaluating spectrum
policies becomes crucial for the sustainability of spectrum management. This
issue is currently under investigation by the National Spectrum Consortium12
(NSC), a research and development organization that incubates new technologies
to tackle challenges about radio spectrum management and utilization.
In this paper, we describe the Dynamic Spectrum Access Policy Framework
(DSA Policy Framework). The DSA Policy Framework supports the manage-
ment of machine-readable radio spectrum usage policies and provides a request
evaluation interface that is able to reason about the policies and generate permit
or deny results to spectrum access requests. This is accomplished via the utiliza-
tion of a novel policy representation based on semantic web standards OWL and
PROV-O that encodes its rules in an ontology. This ontology, combined with
background knowledge originating from a number of relevant select sources, is
stored in a Knowledge Graph that is used by a domain-specific reasoning imple-
mentation that mixes GeoSPARQL [22], OWL reasoning, and knowledge graph
traversal to evaluate policies that are applicable to spectrum access requests.
Effects (Permit/Deny/Permit with Obligations) are assigned and explanations
are provided to justify why a particular request was permitted or denied access
to the requested frequency or frequency range.
DSA Policy Framework 3
2 Sharing the radio spectrum
Radio spectrum policies specify how available spectrum should be used and
shared. The applicability of existing policies must be checked for a variety of
activities that use spectrum in many settings including various different types of
requesters (e.g. systems and devices). For example, training exercises will request
spectrum usage for a certain time frame and geographic region for potentially
hundreds of radios with a wide variety of capabilities. During a training exercise,
local policies are typically created to manage the spectrum that radios used
in the exercise will require, and to minimize interference between federal and
commercial (non-federal) radios that may be operating in the same area and
within the same frequency range. Throughout this paper the following definitions
are used:
High-level policies: Policies as documented by authoritative agencies, includ-
ing NTIA and FCC. These policies are not prone to change in the short term,
although they may evolve when new versions of documents are released.
Local policies : Specializations of high-level policies. These are created to
locally manage the spectrum requests of various devices that want to operate
in specific locations and/or for specific time periods.
Sub-policies: A sub-section of a policy, sometimes referred to as “provisions”
in NTIA Redbook policies.
Spectrum manager: The role of a human who is responsible for managing
policies. The spectrum manager verifies if existing policies are sufficient to
support some activity and creates local policies to accommodate specific
spectrum requirements.
Spectrum system: This role represents some external system that is being
used to generate spectrum requests on behalf of entities that require the use
of a specific frequency or set of frequencies.
3 Dynamic Spectrum Access Policy Framework
The DSA Policy Framework supports the following objectives:
Serve as a centralized machine-readable radio spectrum policy repository.
Provide policy management features (including creation and customization)
for a wide range of radio spectrum domain users.
Use machine-readable policies as a basis for automatically evaluating radio
spectrum access requests.
As shown on the top of Figure 1, the DSA Policy Framework provides two
major functions: Policy Management and Request Evaluation.Policy Manage-
ment enables the spectrum manager to create local policies by referencing rele-
vant higher level policies and adding customization. Request Evaluation utilizes
all policies to automatically process spectrum requests. The results include ref-
erences to any policy that was involved in the evaluation. As policies evolve,
4 H. Santos et al.
the underlying knowledge representation evolves, enabling the request evalua-
tion module to use the most current policy information to reason and assign
effects (permit, deny, permit with obligation) to spectrum requests. Spectrum
managers can verify evaluation results in the presence of newly created policies
through the Request Builder tool.
To support the DSA Policy Framework infrastructure we leveraged Whyis [20],
a nanopublication-based knowledge graph publishing, management, and analy-
sis framework. Whyis enables the creation of domain and data-driven knowledge
graphs. The DSA Policy Framework takes advantage of the use of nanopublica-
tions [15] in Whyis, which allows it to incrementally evolve knowledge graphs
while providing provenance-based justifications and publication credit for each
piece of knowledge in the graph. This is particularly useful for the spectrum
domain because policies do change and new policies can be created, potentially
triggered by multiple sources. The DSA Policy Framework also makes use of the
SETLr [19] Whyis agent, which enables the conversion of a number of the iden-
tified knowledge sources or derivatives to the Resource Description Framework
(RDF), thereby bootstrapping the DSA Knowledge Graph.
Knowledge Sources
Named locations
* shapefiles
NTIA Redbook
* High-level policies
DSA Ontology
* Extracted/Curated terms
Tetherless World
Knowledge Store
Policy management
(Spectrum manager)
Request evaluation
(Spectrum system)
Request Builder
DSA Knowledge Graph
Fig. 1. DSA Policy Framework architecture
3.1 Knowledge Sources
In order to support the creation and interpretation of machine-readable radio
spectrum policies, information from several sources was mined and incorporated
into a Knowledge Graph. Spectrum policies and term definitions were obtained
from the NTIA Redbook, an IEEE Standard Definitions and Concepts docu-
ment [3], and from FCC 14-31 [4] (a FCC policy publication). Specific schemas
associated with the spectrum domain were obtained from the Standard Spectrum
DSA Policy Framework 5
Resource Format (SSRF) [5]. Finally, information about geographical locations
was obtained from the shapefiles.13
During the policy capture process, Rensselaer Polytechnic Institute (RPI)
collaborated with DSA domain experts from Capraro Technologies Inc. and LGS
Labs of CACI International Inc. to select and analyze English text-based policies
from the NTIA Redbook and from various FCC documents. In order to be used
by the DSA Policy Framework, the English text was converted into a different
representation, and many of the terms used in the English text were incorporated
into a domain ontology. During this process, we observed that the text for many
policies is logically equivalent to a conditional expression e.g., IF some device
wants to use a frequency in a particular frequency range AND at a particular
location, THEN it is either PERMITTED or DENIED.
More complex policies contain a set of conditional expressions, with each
conditional expression focused on a particular request attribute, e.g., a request
device type, frequency, frequency range. The spreadsheet displayed in Figure 2
contains an example of a complex policy from the NTIA Redbook called US91.
Due to space constraints, we have omitted some of the sub-policies for US91
and the columns that document policy metadata and provenance, including the
original text, source document, URL, and page number.
RuleID Parsed logical rule Requester Affiliation Frequency Location Effect Obligation
IF RefFreq is ε ( ≥ 1755 MHz
AND ≤ 1780 MHz) then the
following provisions apply
IF TR is AWS AND RefFreq is ε ( ≥
1755 MHz AND ≤ 1780 MHz)
AND (TR has successfully
coordinated on a nationwide basis
prior to operation, unless otherwise
specified by Commision rule, order ,
or notice) THEN TR is Primary
1755 MHz -
1780 MHz
Permit with
IF TR is Federal AND RefFreq is ε (
≥ 1755 MHz AND ≤ 1780 MHz)
AND TR is operating in one the
following locations AND (Until
reaccommodated in accordance with
47 CFR 301) THEN it is Primary
TR Federal
1755 MHz -
1780 MHz
Yuma Proving
Ground, Fort Irwin,
Fort Polk, Fort Bragg,
White Sands Missile
Range, Fort Hood
Permit with
RefFreq is ε ( ≥ 1755 MHz AND ≤
1780 MHz) AND TR is operating in
one the following locations THEN It
is Primary
JTRS Federal
1755 MHz -
1780 MHz
Yuma Proving
Ground, Fort Irwin,
Fort Polk, Fort Bragg,
White Sands Missile
Range, Fort Hood
Permit with
JTRS is Primary
Fig. 2. Spreadsheet excerpt showing the NTIA Redbook US91 policy capture
US91 regulates the usage of the 1755-1780MHz frequency range and is an
example of a spectrum range that must be efficiently shared by both federal and
non-federal devices. Because spectrum usage can vary by device, affiliation and
location, we decompose the policy into several sub-policies (US91-1, US91-2, and
US91-3). A parsed logical rule is manually created for each policy by a domain
expert. The elements of the logical rule are further expressed as attribute-value
6 H. Santos et al.
pairs, e.g., Requester = AWS. The attribute (column) names map into the
following elements of the policy logical expression that is used by the framework:
Requester: the device requesting access
Affiliation: the affiliation of the requester (Federal, Non-Federal)
Frequency: the frequency range or single frequency being requested
Location: location(s) where the policy is applicable
Effect: the effect a policy yields, if the rule is satisfied (Permit, Deny, Permit
with Obligations)
Obligations: the list of obligations the requester needs to comply with in
order to be permitted
The framework utilizes several ontologies to support policy administration
and spectrum request processing, including a domain ontology called the DSA
Ontology. DSA ontology terms were collected during policy capture and/or de-
rived from the NTIA Redbook, IEEE Standards, SSRF, or other domain source.
All terms were curated by an ontology developer and linked to external ontolo-
gies including PROV-O and the Semanticscience Integrated Ontology (SIO) [7].
3.2 Representing radio spectrum requests
Figure 3 shows the DSA request model and a sample request. The model is based
on the World Wide Web Consortium’s recommended standard for provenance
on the web (PROV). The modeling of requests as activities and agents was influ-
enced by the policy attributes described as columns in the policy capture spread-
sheet (shown in Figure 2) and extended to include the action associated with the
requester in a spectrum request (currently we represent only the Transmission
action). In the model, the requester (prov:Agent) is linked with an action
prov:Activity using the prov:wasAssociatedWith predicate. The loca-
tion attribute is represented as prov:Location and linked to the requester us-
ing the prov:atLocation predicate. The time attribute, which describes when
the request action is to start and end, is represented using the literal data type
xsd:dateTime and linked to the action using the prov:startedAtTime and
prov:endedAtTime predicates.
For attributes not natively supported by PROV, we use SIO, which enables us
to model objects and their attributes (roles, measurement values) with the use of
the sio:Attribute class and sio:hasAttribute /sio:hasValue /sio:hasUnit
predicates. In the DSA request model, the attributes Frequency, Frequency
Range, and Affiliation are represented as sio:Attribute and linked to the
requester using sio:hasAttribute. For attributes that can assume a value
(currently only Frequencies), literal values (sio:hasValue) and units of mea-
surement (sio:hasUnit) are used to express the quantification of an attribute
as a specified unit from the Units of Measurement Ontology (UO) [13]. In Fig-
ure 3, a sample radio spectrum request instance is shown, where the requester
is a Generic Joint Tactical Radio System (JTRS) radio. The device is physically
located at a location that is defined by the Well-Known Text (WKT) [2] string
DSA Policy Framework 7
Agent (Requestor) Activity (Action)
hasAttribute 2019-05-01T01:00:00Z
(-114.23 33.20)
Fig. 3. The DSA request model and sample request
POINT(-114.23 33.20) and is requesting access to the frequency range 1755
- 1756.25 MHz using the FrequencyRange attribute, which is described
by the composition of the FrequencyMinimum and FrequencyMaximum at-
tributes, with their respective value and unit.
3.3 Representing radio spectrum policies
The DSA policy model was created to enable (1) the unambiguous representation
of rules as parsed during policy capture, (2) the reuse of an existing policy’s rules
for creating local policies, and (3) the implementation of request evaluation
capabilities. The model is based on OWL, encoding policy rules as restrictions
in OWL classes that represent policies. The OWL restrictions are constructed
on the RDF properties of the DSA request model, presented in the previous
subsection, while constraining their ranges to the expected literal value or class,
as dictated by the policy rule being expressed. To demonstrate this, Listing 1.1
shows the OWL expression in Manchester syntax of an excerpt of the NTIA US91
policy that describes how a requester that is an example of a radio, categorized
as JTRS, can use the spectrum regulated by US91 (1755-1780 MHz) in specific
locations (US91-3 sub-policy in Figure 2).
The model advocates for the creation of an OWL class for each rule that
composes the complete policy rule expression. In the example, the OWL class
US91 has the restriction on the frequency range attribute, with the minimum
value 1755 and maximum value 1780 (lines 4-11). This class is a subclass of
Transmission (line 13), which is the action this policy regulates. In lines 15-
20, a class US91-3 restricts the requester to a JointTacticalRadioSystem
and appends that rule to the rules of its super-class US91. The last class of this
policy expression is US91-3.1, which encodes the location rule as a restriction
on the atLocation property (lines 24-25), constraining it to a location class.
When the composition of rules expressed by this last class is evaluated to be
true, the policy should assign the permit effect. This is expressed by asserting
US91-3.1 as a subclass of Permit (line 27).
8 H. Santos et al.
1Class: US91
3 Transmission and
4 (wasAssociatedWith some (hasAttribute some
5 (FrequencyRange
6and (hasAttribute some
7 (FrequencyMaximum and
8 (hasValue some xsd:float[<= 1780.0f])))
9and (hasAttribute some
10 (FrequencyMinimum and
11 (hasValue some xsd:float[>= 1755.0f]))))))
12 SubClassOf:
13 Transmission
15 Class: US91-3
16 EquivalentTo:
17 US91 and
18 (wasAssociatedWith some JointTacticalRadioSystem)
19 SubClassOf:
20 US91
22 Class: US91-3.1
23 EquivalentTo:
24 US91-3 and (wasAssociatedWith some
25 (atLocation some US91-3.1_Location))
26 SubClassOf:
27 Permit, US91-3
Listing 1.1. OWL expression of part of the US91 policy in Manchester syntax
Many of the NTIA policies, including US91, contain location rules and, often,
these rules are written in terms of location lists. To represent this, we defined
OWL classes for the lists to be used in conjunction with the OWL expression
of policies. Listing 1.2 shows the definition of the US91-3.1 Location class,
which is used in the US91-3.1 restriction, as previously shown. We have used
the GeoSPARQL predicate sfWithin in conjunction with OWL unions to ex-
press that the rule is satisfied if the location is specified in the list (lines 3-8).
In this example, we leverage location information we imported from
shapes for US Federal locations.
1Class: US91-3.1_Location
3 (sfWithin value White_Sands_Missile_Range) or
4 (sfWithin value Ft_Irwin) or
5 (sfWithin value Yuma_Proving_Ground) or
6 (sfWithin value Ft_Polk) or
7 (sfWithin value Ft_Bragg) or
8 (sfWithin value Ft_Hood)
10 Location
Listing 1.2. OWL expression of a location list in Manchester syntax
DSA Policy Framework 9
The creation of the OWL class hierarchy (and, therefore, the incremental
addition of rules) maximizes the reuse of rules when spectrum managers create
local policies. To illustrate this claim, Listing 1.3 shows a sample local policy
which modifies the existing US91-3 sub-policy to Deny requests in a specific
time window. Lines 4-5 contain the time restrictions on the PROV-O predicates.
Local policies can have an explicit precedence level, as shown in line 7.
1Class: US91-3.1-Local
3 US91-3.1 and
4 endedAtTime some xsd:dateTime[>=2019-10-01T11:00:00Z] and
5 startedAtTime some xsd:dateTime[<=2019-10-01T17:00:00Z]
7 Deny, Priority_1, US91-3.1
Listing 1.3. OWL expression of a local policy in Manchester syntax
3.4 Policy management
The DSA Policy Framework provides a web interface to allow spectrum managers
to have a comprehensive understanding of the DSA Knowledge Graph, includ-
ing policies, locations, and entities in the DSA Ontology. It leverages Whyis
default “views” with some extensions for supporting a domain-specific display
of pieces of the Knowledge Graph. The structure and content of the interface are
driven by the DSA Knowledge Graph, which ensures that it displays relevant
and contextualized information and features.
The Policy Faceted Browser that allows a user to quickly find policies based
on the selection of attribute values. Spectrum managers can, for instance, find
policies applicable to a list of select locations or policies applicable to a spe-
cific device, or even to a combination of both. This is accomplished by domain-
specific SPARQL queries that retrieve and group attributes from the policy’s
OWL structure. The user can view details of a policy or reuse a policy’s rules.
The Policy Detail view provides a display of policy metadata, including name,
original text and identifier, and a human-readable version of the policy encoded
rules. If the policy specifies locations, those locations will be displayed on a map.
The Policy Builder view can be used to build a policy from scratch or to cre-
ate local policies by reusing existing policies’ rules. The Policy Builder leverages
the DSA Knowledge Graph to provide user support during policy creation. For
instance, if a user wants to create a rule for an specific device, the Builder will
display known devices as represented in the KG. More than that, the Builder
“understands” the semantics of the rule which means that if the user wants to
add a frequency range rule, for instance, the Builder will prompt the user to enter
both lower and upper bound values. Users can also set policy effects, precedence,
and obligations. In the back end, the policy is converted to the DSA policy model
in OWL and stored as a new piece of knowledge in the DSA Knowledge Graph.
Knowledge curation is a time-consuming task and it might impact the pace at
which updated knowledge becomes available for use when creating new policies.
10 H. Santos et al.
To overcome this, the DSA Policy Framework supports policy additions that
refer to terms not currently in the ontology, by allowing input of new terms
along with the annotation that those terms need in order to be curated by the
appropriate ontology owner. This allows a spectrum manager to input and test
a new policy without having to wait for an ontology update first.
3.5 Request evaluation: Domain-specific reasoning
To enable the evaluation of spectrum requests against policies, we have imple-
mented a domain-specific reasoner that combines various Semantic Web compu-
tational approaches to assign Permit/Deny effects to requests, while fulfilling
requirements, including geographical reasoning, policy precedence evaluation,
and explanations for denied requests. The domain-specific reasoner follows a
four-phase pipeline, with a set of requests as input and the assigned effect, a
list of obligations, and a list of reasons for each request as output. The reasoner
initiates by creating an in-memory RDF graph originated by the merge of the
request RDF graph and the DSA Knowledge Graph.
Next, in the geographical reasoning phase, the reasoner elicits the ge-
ographical relationships among the requests’ WKT locations, and the named
locations present in the DSA KG, by using GeoSPARQL to infer triples like
:req location geo:sfWithin :NAMED LOCATION. The inferred triples are
then asserted back into the graph.
In the OWL reasoning phase, the reasoner makes use of the HermiT OWL
reasoner [14] to perform Description Logic (DL) reasoning over the updated
graph. The domain-specific implementation relies on the DL services of:
Classification for computing all subclass relationships, allowing the inferred
class hierarchy to be leveraged when querying policy Effect and Precedence.
Realization of computing classes (policies) that individuals (requests) belong
to. “Belonging” to a class means that an individual satisfies the constraints of
that class; realization can be understood as determining policy applicability.
DL Query for retrieving the individuals that were determined to belong to
specific classes.
Using the list of applicable policies retrieved from HermiT, the domain-
specific reasoner decides precedence, in the precedence evaluation phase,
by a simple evaluation of which policy has the highest precedence level. Poli-
cies with no explicit precedence are assumed to have the lowest precedence. The
highest precedence policy effect is then assigned to the request. The reasoner
follows these heuristics to explain the assignment of a Deny effect to a request,
in the explanation phase:
Requests can be denied by a specific applicable policy with a Deny effect.
In this case, the “reason” for the denial is the identity of this policy and
the specification of what attributes of the request fulfill the rules contained
within. The reasoner retrieves the policy’s rules and presents them as reasons
for denial.
DSA Policy Framework 11
Requests can be denied due to a lack of a policy with a Permit effect (i.e.
there is no applicable policy with either a Deny or Permit effect). In these
cases the reasoner finds the rules that the request did not satisfy in order to
be assigned a Permit. The reasoner calculates the paths in the OWL class
hierarchy from those policies that the request was reasoned to belong to, to
the policies that would result in a Permit. These paths contain classes that
the request was determined not to belong to. Unfulfilled rules for each policy
found to be in the path are retrieved and presented as reasons for the Deny.
As as example, the request in Figure 3 would ultimately be determined to
belong to the US91,US91-3, and US91-3.1 classes displayed in Listing 1.1
and, therefore, assigned the Permit effect. Nevertheless, when the local policy
shown in Listing 1.3 exists in the graph, the request is then reasoned to belong
to the US91-3.1-Local class as well. Since this local policy contains a time
constraint rule with a higher precedence, the reasoner assigns the effect to be a
Deny and returns the reason, “the request is in a prohibited time window.”
Conversely, if the request is modified to a different location outside the lo-
cations expressed in Listing 1.2, it would be reasoned to belong only to US91
and US91-3 classes, based on the frequency range and requester attributes of
the request. These applicable classes don’t express an explicit Permit or Deny
effect, so the reasoner must default to Deny. The reasoner then calculates the
path to the class US91-3.1 that would Permit it if the conditions had been
met, and retrieves the rules of each class in the path (in this case, only the
rules of the US91-3.1 class) and returns the reason, “the request is not in a
permitted location.”
4 Evaluation of the Framework
In order to evaluate the DSA Policy Framework’s semantic representation of the
domain, we identified several fundamental spectrum policy constructs. They are
displayed in bold in the first column of Table 1. For each of them, we worked with
domain experts to identify the elements that were required in order to effectively
represent policies and support request evaluation. The table contains columns
for Policy Representation (high-level and local policies) and Request Evaluation.
“Yes” in the columns indicates that the policy construct is either Relevant or it
has been fully addressed and Implemented. “Partial” indicates that the current
implementation meets a simplified version of the requirement, while “uc” means
that the construct element is currently under consideration.
The basic structure of a policy varies among source documents. The NTIA
Redbook uses provisions to distinguish policy behavior with regards to attributes
(specific device or location, for instance). Provisions are described as sub-policies
in the policy capture spreadsheet and in the Framework, and they are leveraged
during request evaluation. Policies can be specified in the Parser logical rule
format, which is enabled in the Policy Builder. The framework does not currently
represent the individual elements of an obligation. Instead, it is represented as
text or as a canned identifier to support request evaluation.
12 H. Santos et al.
Policy Representation Request Evaluation
Domain Policy Construct Relevant Implemented Relevant Implemented
Provision yes yes yes yes
Parsed logical rule yes yes no no
Obligation yes partial yes partial
Device yes yes yes yes
Organization yes yes yes yes
Dependency yes uc uc uc
Licensee yes partial yes partial
Federal/Non-Federal yes partial yes partial
Named requester yes yes yes yes
Frequency range yes yes yes yes
Single frequency yes yes yes yes
Named bands yes yes uc uc
Units yes yes yes yes
Timezones yes yes yes yes
Policy validity yes yes yes yes
Named locations yes yes yes yes
Relative locations yes uc yes uc
Polygons/Circles yes yes yes yes
Geographical rules
Specific location yes yes yes yes
List of locations yes yes yes yes
Levels yes yes yes yes
Policy triggered yes yes yes yes
Rules not satisfied yes yes yes yes
Table 1. DSA Policy Framework policy semantics coverage
The text in the source policies specifies regulations for a variety of requester
types. While all requester types are relevant for policy representation, requests
only express the device. However, the DSA KG does represent several types
including Organization and Licensee, so if a request was received for one of
these types, theoretically, it could be evaluated. If a policy specifies a dependency
between requesters, this dependency is only currently treated as text. Policies can
also regulate the spectrum usage by affiliation (Federal systems are permitted
to use some frequency, for instance). The Framework allows affiliation rules to
be specified in policies, however, affiliation reasoning is currently limited by the
expressiveness of affiliations in the DSA Ontology, where only requesters with
an explicit affiliation (named requester) are effectively reasoned.
Frequency rules are specified in terms of a range or single frequency in dif-
ferent units (MHz, GHz). Sometimes, a range is described as a band, which is a
named frequency range. Requests do not express named bands. The framework
supports all of these constructs, using the Frequency and FrequencyRange
attributes, and the sio:hasUnit property. The time attribute exists in local
DSA Policy Framework 13
policies only, and they specify the validity of a policy. Time and timezones are
supported using xsd:dateTime literals and PROV-O time predicates.
Most policies refer to locations by names or by coordinates (points, poly-
gons, and circles), but sometimes a location is expressed in relation to another
location. The framework uses shapes to refine named locations and
WKT literals to represent polygons and circles. Relative locations are still un-
der development. Geographical rules are defined in terms of the requester being
in a location or in a list of locations. The framework implements all of these
constructs using geo:sfWithin predicate and OWL unions.
The framework implements all of the identified precedence needs. It can de-
fine and evaluate precedence levels. For explaining evaluation results, the frame-
work implements all current explanation requirements. It outputs which policy
was triggered by a request and presents reasons based on the presented heuris-
tics. Finally, the constructs identified as “uc” are areas for future work.
5 System Adoption and Deployment
The DSA Policy Framework is being used in simulated scenarios, where it sup-
ports the research & development of other components of a dynamic spectrum
management system. It currently contains approximately 165 high-level policies
from the NTIA Redbook (including their sub-policies). The DSA Ontology con-
tains 695 classes and is constantly evolving to address new domain constructs
and support more precise request evaluation.
The framework is transitioning to support live, over-the-air field exercises
involving a diverse set of federal and commercial radios. During these exercises,
the Framework supports (1) the creation, deletion, and revision of local policies,
(2) the real-time processing of numerous spectrum requests, and (3) the gener-
ation of explanations that describe how the spectrum requests were processed.
The public released assets developed during the course of the project can be
accessed at
6 Related Work
Kirrane [17] offers a comprehensive survey of access control models, well-known
policy languages, proposed frameworks that utilize ontologies and/or rules to ex-
press policies, and a categorization of policy languages and frameworks against
access control requirements. XACML 3.0, the eXtensible Access Control Markup
Language [1], is a well-known policy language and de facto standard for repre-
senting attribute-based access control (ABAC) [16] policies and requests. Impor-
tantly, XACML provides a reference architecture for centralizing access control
and a process model for evaluating requests against existing policies that inform
the design of access control systems across domains and technologies.
Thi [24] proposes an OWL-based extension to XACML to support a gener-
alized context-aware role-based access control (RBAC) model providing spatio-
14 H. Santos et al.
temporal restrictions and conforming with the NIST RBAC standard [9]. Their
work augments the XACML architecture with new functions and data types.
Muppavarapu [21] identifies the limitations of identity-based access control
schemes used in the Open Grid Services Architecture (OGSA) and proposes the
use of OWL to represent the ontology of an organization’s resources and users.
They further propose the use of semantics in conjunction with the XACML
standard for better interoperability and reduced administration overhead.
Our approach combines OWL, PROV-O, and the HermiT OWL reasoner
with an ontology, represented as a knowledge graph, to support the representa-
tion of policies governing access to available spectrum. Relevant related research
is described in Dundua [8], where previous work is described that uses OWL for
modeling and analyzing access control policies, especially ABAC, and considers
how the ABAC model can be integrated into ontology languages. In addition,
Sharma [23] describes how OWL can be used to formally define and process secu-
rity policies that can be captured using ABAC models. This work demonstrates
how models, domains, data and security policies can be expressed in OWL and
how a reasoner can be used to decide what is permitted.
Kolovski [18] maps the web service policy language, WS-Policy [6], to the
description logic fragment species of OWL and demonstrates how standard OWL
reasoners can check policy conformity and perform policy analysis tasks.
Garc`ıa [12,11] and Finin [10] offer important contributions on how end-to-end
usage rights and access control systems may be implemented in OWL and RDF.
Garc`ıa proposes a “Copyright Ontology” based on OWL and RDF for expressing
rights, representations that can be associated with media fragments in a web-
scale “rights value change.” Finin describes two ways to support standard RBAC
models in OWL and discusses how their OWL implementations can be extended
to model attribute-based RBAC or, more generally, ABAC.
Our policy representation approach builds on previous work by matching
the cross-domain policy expression semantics of XACML. It extends these se-
mantics with the capacity to express rich spatio-temporal restrictions, enabling
the implementation of a wide variety of attribute-based policies across domains.
It leverages background knowledge from domain-specific knowledge graphs that
are structured with a domain-derived ontology, enabling the inference of pol-
icy applicability based on attributes and constraints. Our approach uniquely
conceptualizes policy requests as PROV activities and request evaluations as
realizations. Finally, our approach provides a novel reasoner-based explanation
in request evaluation results, enabling domain policy developers to understand
the precise reasons for policy decisions.
7 Conclusion
We described a policy Framework that leverages a machine-readable, radio spec-
trum policy representation to support policy management and enable a domain-
specific reasoner to efficiently process spectrum requests. The DSA policy model
uses OWL restrictions on PROV-O properties to represent policy rules in a hi-
DSA Policy Framework 15
erarchical approach that maximizes the reuse of rules when local policies are
created, therefore facilitating the creation of local policies. The hierarchical na-
ture of this policy representation also supports the explanation of evaluation
results, by traversing the graph to find rules that were not satisfied. Because it
leverages a domain Knowledge Graph, built from multiple knowledge sources, the
domain-specific reasoner allows rich semantics which otherwise would be difficult
to achieve with approaches that rely on a “flat” representation of attributes.
During the course of the project, we encountered OWL reasoning perfor-
mance issues when multiple requests are simultaneously received. To decrease
reasoning time, we partitioned the request graph into smaller graphs and eval-
uated each in parallel, using multiple processor cores. This allowed us to match
the required response time of under 10 seconds.
Future work includes additional support for enforcement of the DSA policy
model. Although the DSA policy model’s OWL hierarchy maximizes the reuse of
rules, there is currently no enforcement. Overlapping rules can be created, which
can lead to multiple policies with the same precedence level being triggered dur-
ing request evaluation. The DSA Ontology is constantly changing as additional
policies are added to the framework with terms that have yet to be defined.
Existing policies can be affected by changes in the DSA Ontology as their rules
reference entities in it and they may need to be reviewed with regards to the
updated representation of the domain. We plan to provide spectrum managers
a way of tracking policies that are subject to review due to ontology changes.
Acknowledgement of Support and Disclaimer. This work is funded in
support of National Spectrum Consortium (NSC) project number NSC-17-7030.
Any opinions, findings and conclusions or recommendations expressed in this
material are those the authors and do not necessarily reflect the views of AFRL.
1. eXtensible Access Control Markup Language (XACML) Version 3.0. http:// 3.0-core- spec-os- en.html
2. ISO/IEC 13249-3:2016 Information technology — Database languages — SQL mul-
timedia and application packages — Part 3: Spatial
3. IEEE Standard Definitions and Concepts for Dynamic Spectrum Access: Terminol-
ogy Relating to Emerging Wireless Networks, System Functionality, and Spectrum
Management. IEEE Std 1900.1-2008 pp. 1–62 (2008)
4. Federal Communications Commission, FCC 14-31. Tech. rep. (2014)
5. Standard Spectrum Resource Format (SSRF), Data Exchange Standard, Version
3.1.0 (MC4EB Pub 8). Tech. rep. (2014)
6. Curbera, F., Hallam-Baker, P., Hondo, V.M., Nadalin, A., Nagaratnam, N., Sharp,
C.: Web services policy framework (ws-policy) (2006),
7. Dumontier, M., Baker, C.J., Baran, J., Callahan, A., Chepelev, L., Cruz-Toledo,
J., Del Rio, N.R., Duck, G., Furlong, L.I., Keath, N., Klassen, D., McCusker,
J.P., Queralt-Rosinach, N., Samwald, M., Villanueva-Rosales, N., Wilkinson, M.D.,
Hoehndorf, R.: The Semanticscience Integrated Ontology (SIO) for biomedical re-
search and knowledge discovery. Journal of Biomedical Semantics 5, 14 (2014)
16 H. Santos et al.
8. Dundua, B., Rukhaia, M.: Towards Integrating Attribute-Based Access Control
into Ontologies. In: 2019 IEEE 2nd Ukraine Conference on Electrical and Computer
Engineering (UKRCON). pp. 1052–1056 (2019)
9. Ferraiolo, D.F., Kuhn, D.R.: Role-based access controls (2009)
10. Finin, T., Joshi, A., Kagal, L., Niu, J., Sandhu, R., Winsborough, W., Thurais-
ingham, B.: ROWLBAC: representing role based access control in OWL. In: Pro-
ceedings of the 13th ACM symposium on Access control models and technologies.
pp. 73–82. SACMAT ’08, Association for Computing Machinery, Estes Park, CO,
USA (2008)
11. Garc´ıa, R., Castell`a, D., Gil, R.: Semantic copyright management of media frag-
ments. In: DATA 2013 - Proceedings of the 2nd International Conference on Data
Technologies and Applications. pp. 230–237 (2013)
12. Garc´ıa, R., Gil, R., Delgado, J.: A web ontologies framework for digital rights
management. Artificial Intelligence and Law 15(2), 137–154 (2007)
13. Gkoutos, G.V., Schofield, P.N., Hoehndorf, R.: The Units Ontology: a tool for
integrating units of measurement in science. Database 2012 (2012)
14. Glimm, B., Horrocks, I., Motik, B., Stoilos, G., Wang, Z.: HermiT: An OWL 2
Reasoner. Journal of Automated Reasoning 53(3), 245–269 (2014)
15. Groth, P., Gibson, A., Velterop, J.: The anatomy of a nanopublication. Information
Services & Use 30(1-2), 51–56 (2010), publisher: IOS Press
16. Hu, V.C., Kuhn, D.R., Ferraiolo, D.F., Voas, J.: Attribute-based access control.
Computer 48(2), 85–88 (2015)
17. Kirrane, S., Mileo, A., Decker, S.: Access control and the Resource Description
Framework: A survey. Semantic Web 8(2), 311–352 (2017)
18. Kolovski, V., Parsia, B., Katz, Y., Hendler, J.: Representing Web Service Policies
in OWL-DL. In: Gil, Y., Motta, E., Benjamins, V.R., Musen, M.A. (eds.) The
Semantic Web – ISWC 2005. pp. 461–475. Lecture Notes in Computer Science,
Springer, Berlin, Heidelberg (2005)
19. McCusker, J.P., Chastain, K., Rashid, S., Norris, S., McGuinness, D.L.: SETLr:
the semantic extract, transform, and load-r. Tech. Rep. e26476v1, PeerJ Inc. (2018)
20. McCusker, J., Rashid, S.M., Agu, N., Bennett, K.P., McGuinness, D.L.: The Whyis
Knowledge Graph Framework in Action. In: International Semantic Web Confer-
ence (P&D/Industry/BlueSky) (2018)
21. Muppavarapu, V., Chung, S.M.: Semantic-Based Access Control for Grid Data Re-
sources in Open Grid Services Architecture - Data Access and Integration (OGSA-
DAI). In: 2008 20th IEEE International Conference on Tools with Artificial Intel-
ligence. vol. 2, pp. 315–322 (2008), iSSN: 2375-0197
22. Perry, M., Herring, J.: OGC GeoSPARQL-A geographic query language for RDF
data. OGC implementation standard 40 (2012)
23. Sharma, N.K., Joshi, A.: Representing Attribute Based Access Control Policies
in OWL. In: 2016 IEEE Tenth International Conference on Semantic Computing
(ICSC). pp. 333–336 (2016)
24. Tran Thi, Q.N., Dang, T.K.: X-STROWL: A generalized extension of XACML for
context-aware spatio-temporal RBAC model with OWL. In: Seventh International
Conference on Digital Information Management (ICDIM 2012). pp. 253–258 (2012)
25. Zhang, Y.: Dynamic Spectrum Access in Cognitive Radio Wireless Networks. In:
2008 IEEE International Conference on Communications. pp. 4927–4932 (2008),
iSSN: 1938-1883
26. Zhao, Q., Sadler, B.M.: A Survey of Dynamic Spectrum Access. IEEE Signal Pro-
cessing Magazine 24(3), 79–89 (2007)
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Attribute-based access control (ABAC) is a flexible approach that can implement AC policies limited only by the computational language and the richness of the available attributes, making it ideal for many distributed or rapidly changing environments.
Conference Paper
Full-text available
The rapid growth of location-based applications, geographic or large scale information systems has resulted in the demand of strictly controlling data access that requires specifying and enforcing fine grained policies with the variety of context-aware spatial and temporal restrictions. Besides, the interoperable use of distributed and heterogeneous data such as data sharing, data integration or data exchanging between different organizations has caused the formation and development of access control mechanisms using XML for enforcing security rules and policies in accordance with the international standards. In this paper, we propose an extension of XACML called the X-STROWL model for a generalized context-aware role-based access control (RBAC) model with the support of spatio-temporal restrictions and in conformity with the NIST standard for RBAC. In doing this, the XACML architecture is augmented with new functions and data types. In addition, policies are reorganized to adopt with the NIST standard. Besides, a set of conditions aimed to a certain circumstance can be generalized into a context profile and specified in the access control policies. The model also integrates the OWL ontology for semantic reasoning on hierarchical roles to simplify the specification of access control policies and increase the intelligence of the authorization decision engine.
Full-text available
The Semanticscience Integrated Ontology (SIO) is an ontology to facilitate biomedical knowledge discovery. SIO features a simple upper level comprised of essential types and relations for the rich description of arbitrary (real, hypothesized, virtual, fictional) objects, processes and their attributes. SIO specifies simple design patterns to describe and associate qualities, capabilities, functions, quantities, and informational entities including textual, geometrical, and mathematical entities, and provides specific extensions in the domains of chemistry, biology, biochemistry, and bioinformatics. SIO provides an ontological foundation for the Bio2RDF linked data for the life sciences project and is used for semantic integration and discovery for SADI-based semantic web services. SIO is freely available to all users under a creative commons by attribution license. See website for further information: .
Full-text available
Units are basic scientific tools that render meaning to numerical data. Their standardization and formalization caters for the report, exchange, process, reproducibility and integration of quantitative measurements. Ontologies are means that facilitate the integration of data and knowledge allowing interoperability and semantic information processing between diverse biomedical resources and domains. Here, we present the Units Ontology (UO), an ontology currently being used in many scientific resources for the standardized description of units of measurements.
In recent years we have seen significant advances in the technology used to both publish and consume structured data using the existing web infrastructure, commonly referred to as the Linked Data Web. However, in order to support the next generation of e-business applications on top of Linked Data suitable forms of access control need to be put in place. This paper provides an overview of the various access control models, standards and policy languages, and the different access control enforcement strategies for the Resource Description Framework (the data model underpinning the Linked Data Web). A set of access control requirements that can be used to categorise existing access control strategies is proposed and a number of challenges that still need to be overcome are identified.
This system description paper introduces the OWL 2 reasoner HermiT. The reasoner is fully compliant with the OWL 2 Direct Semantics as standardised by the World Wide Web Consortium (W3C). HermiT is based on the hypertableau calculus, and it supports a wide range of standard and novel optimisations that improve the performance of reasoning on real-world ontologies. Apart from the standard OWL 2 reasoning task of entailment checking, HermiT supports several specialised reasoning services such as class and property classification, as well as a range of features outside the OWL 2 standard such as DL-safe rules, SPARQL queries, and description graphs. We discuss the system’s architecture, and we present an overview of the techniques used to support the mentioned reasoning tasks. We further compare the performance of reasoning in HermiT with that of FaCT++ and Pellet—two other popular and widely used OWL 2 reasoners.
Conference Paper
In this paper, we propose a semantic-based access control method for grid data resources in the open grid services architecture - data access and integration (OGSA-DAI). The OGSA-DAI is an efficient grid-enabled middleware implementation of interfaces and services to access and control data resources (such as files, relational databases and XML databases). However, the identity-based access control in the OGSA-DAI causes substantial overhead for the resource providers in virtual organizations (VOs), because the access control information of individual users has to be maintained by each resource provider. In addition, access control policies need to be specified and managed across multiple VOs. To solve these problems, we propose the use of semantic-based access control policies in data grids. We use the Web ontology language (OWL) standard to represent the ontology of an organization's resources and users. Recently, eXtensible Access Control Markup Language (XACML) has been increasingly used for the representation of access control policies in grid environments. We propose the use of semantics in conjunction with the XACML standard for better interoperability and reduced administration overhead.
Conference Paper
Cognitive radio wireless networks is an emerging communication paradigm to effectively address spectrum scarcity challenge. Spectrum sharing enables the secondary unlicensed system to dynamically access the licensed frequency bands in the primary system without any modification to the devices, terminals, services and networks in the primary system. In this paper, we propose and analyze new dynamic spectrum access schemes in the absence or presence of buffering mechanism for the cognitive secondary subscriber (SU). A Markov approach is developed to analyze the proposed spectrum sharing policies with generalized bandwidth size in both primary system and secondary system. Performance metrics for SU are developed with respect to blocking probability, interrupted probability, forced termination probability, non-completion probability and waiting time. Numerical examples are presented to explore the impact of key systems parameters like the traffic load on the performance metrics. Comparison results indicate that the buffer is able to significantly reduce the SU blocking probability and non-completion probability with very minor increased forced termination probability. The analytic model has been verified by extensive simulation.