IEEE Circuits and Systems Magazine - Q3 2019 - 35

Abstract
This article surveys the state of the art in spectrum prediction
and learning, summarizes applications, techniques, main metrics, computational complexity, and provides practical examples.
We focus on a cellular case study and define required improvements to database-assisted spectrum sharing. The use of history
information and predictive spectrum modeling at different time
scales provides valuable information to regulators, operators,
and users of dynamic spectrum access networks. Prediction enables dynamic spectrum sharing systems to operate proactively,
and consequently improves the performance in terms of reducing
delays and interference among coexisting systems. Current database-assisted spectrum sharing concepts are in fact too static for
many applications. Our numerical results on local-aware predictive spectrum allocation show the advantage of predictive operation in a vehicle-to-everything (V2X) scenario.

Introduction
pectrum sharing technologies have advanced significantly since the concept of cognitive radio (CR)
was proposed in the late 1990s. Dynamic spectrum
access (DSA) technologies, already included in multiple
standards in different frequency bands and regulatory
processes, are being updated to include new forms of licensing. In primary-secondary sharing scenarios, knowledge of primary users' (PU) occupancy distribution helps

minimizing switching latencies and interference. It also
helps in selecting channels to sense and optimize the
sensing order as well as using the most promising channels. Therefore, an intelligent DSA system should have the
ability to estimate the PU traffic patterns, classify them
according to their distributions, and predict the future opportunities using this information [1]-[2] appropriately in
an uncertain environment. This enables the selection of
the best available radio access techniques (RATs) and frequency bands based on application requirements, determining priority and support for the communication needs
of all users.
Information about spectrum use over time, frequency,
and spatial domains provides valuable knowledge to regulators and decision makers about the efficiency of the current use of spectrum allocations - and helps in making informed decisions on future spectrum assignments [3]. To
this end, learning, spectrum prediction, traffic classification, and mobility prediction have been intensively studied
in recent years [4]-[6], and various techniques and applications have been identified. Concurrently, database (DB)
assisted spectrum sharing concepts have been proposed [7]. Due to the introduction of small cells and very
dense fifth generation (5G) systems, in which spectrum sharing

Extent of Sharing

Exclusive Spectrum
Licensing

Licensed Shared
Access

License Exempt
Shared Access

License Exempt
Spectrum

A Single Primary
User for Each
Spectrum Block

A Single Incumbent
User. Licensed Use
of Spectrum by a
Secondary User (SU)

A Single Primary
User of Spectrum.
SUs Allowed to Use
Spectrum in
Restricted Scenarios

Spectrum Commons
Without a Primary
User

Spectrum for Typical
Mobile Network
Operator

Considered for the
Release of Some
Spectrum, e.g., 2.3 GHz
Band, 3.6 GHz Band

Cognitive Radios,
TV White Space
Operation

Industrial, Scientific,
and Medical (ISM)
Spectrum for Wi-Fi,
Bluetooth, LTEUnlicensed etc.

Guarantee of Protection Against Interference

Figure 1. Paradigms of 5G spectrum use.

THIRD QUARTER 2019

IEEE CIRCUITS AND SYSTEMS MAGAZINE

IEEE Circuits and Systems Magazine - Q3 2019

Table of Contents for the Digital Edition of IEEE Circuits and Systems Magazine - Q3 2019