IEEE Circuits and Systems Magazine - Q3 2019 - 41

Ability to predict the appearance of the users in the
spectrum, based on their previous spectrum use and
mobility, would help in timing the spectrum allocations,
performing intelligent handovers, and using the spectrum more efficiently. Table 2 defines a non-exhaustive
list of examples how technical advances and prediction
could be applied in improving the performance of database-assisted spectrum sharing where the mobile network is the secondary user of the spectrum. Most of the
ideas are applicable to both CBRS and LSA systems. The
analysis is generally applicable to any frequency band
where databases are used to enable spectrum sharing.
Predictive channel selection enables, e.g., minimization
of the number of channel changes, which is important
especially in the cellular scenario where a single frequency change takes significant amount of time. One of
the most important development paths to enable fast dynamic adaptations in cellular spectrum sharing systems
is development of equipment and procedures to support DSA operations.

Developed database approaches such as LSA operation in the 2.3 GHz band and CBRS operation in the
3.5 GHz band have focused on sharing the spectrum in
the spatial domain. Sharing in time domain has been
quite limited due to practical limitations in the current
cellular equipment that are designed for more static frequency use. The proposed procedures work quite well
in the developed bands [8, 36]. However, the spectrum
use in many cases is more efficient when the operation can be made more dynamic. The spectrum sharing concept and procedures have to be optimized to
different bands according to the characteristics of users sharing that band. Sharing the band with radars,
satellites, or broadcasting requires bespoke procedures. If the spectrum is shared with mobile satellites,
one can predict where and when interference would
occur according to the movement and coverage area
of the satellites [37]. The database should incorporate
information such as orbital parameters to determine
ephremides, spectrum use, locations of gateways etc.

Table 2.
Proposed improvements to spectrum sharing in database-assisted scenarios.
Proposed change

Status now

Target

Optimization of base stations to
support fast frequency changes

Base stations developed with static
frequency use in mind

Fast frequency changes, significant
delay reduction and improvement in
total throughput

Enabling frequency change on the
fly without shutting down the base
stations

E.g., in the CBRS system one needs
to lock the base station during the
evacuation process

Changing immediately to other available
(proactively selected) channel. Reduction
of the total evacuation and frequency
change time to less than half [8]

Inclusion of traffic-aware predictive
channel selection method to the
system

History information not used
efficiently or not at all in LSA or
CBRS systems

Ability to find best channels for each
application especially when multiple
possibilities available

Joint mobility and spectrum prediction
to support resource management

Mobility management entity (MME)
responsible for seamless handovers.
QoS support limited

Spatial database that includes both actual
spectrum use data and measured QoS
values as inputs for MME

Modifying LSA to allow more dynamic
use of resources

Currently resource blocks are
dedicated to certain licensees

Flexible use, pool of resource blocks
among LSA licensees instead of
dedicated ones

Use of the SAS and LSA concepts in
new bands such as 26 GHz

Satellite use, 26 GHz recommended
as a "pioneer 5G band" in Europe

Sharing band with satellites, predictive
local-aware allocations

Predictive network slicing according
to QoS requirements, prioritization of
traffic

Slicing under development, network
resources underused

Flexible allocation of frequencies and
RATs for each slice

Deep learning for long term spectrum
analysis

Long term history data not used
efficiently in spectrum management

Intelligent regulatory decisions and
spectrum allocations based on the
learned data

Use of interference maps to enable
predictive resource allocation across
time, frequency, and spatial domains

Spectrum measurements and
resource allocations decisions made
locally

Both high-end devices and crowdsourcing used in collecting spectrum
data over the area of interest
continuously

THIRD QUARTER 2019

IEEE CIRCUITS AND SYSTEMS MAGAZINE

41
IEEE Circuits and Systems Magazine - Q3 2019

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