Adaptive power and frequency allocation strategies in cognitive radio systems

Dissertation

Research output: ThesisDissertationCollection of Articles

Abstract

This doctoral thesis comprises a summary of novel results considering (1) channel selection in a cognitive radio system (CRS) using history information and (2) power allocation in a selected frequency band assuming a fading channel. Both can be seen as methods to manage interference between in-system users as well as to the users of other systems operating in the same geographical area and frequency band. Realization of CRSs that are using various methods to obtain information about environment and making intelligent decisions based on that information requires the use of adaptive transmission. Adaptive techniques proposed in this thesis enable efficient operation of CRSs in varying radio environment. History information and learning are essential factors to consider in the CRS design. Intelligent use of history information affects throughput, collisions and delays since it helps to guide the sensing and channel selection processes. In contrast to majority of approaches presented in the literature, this thesis proposes a classification-based prediction method that is not restricted to a certain type of traffic. Instead, it is a general method that is applicable to a variety of traffic classes. The work develops an optimal prediction rule for deterministic traffic pattern and maximum likelihood prediction rule for exponentially distributed traffic patterns for finding channels offering the longest idle periods for secondary operation. Series of simulations were conducted to show the general applicability of the rule to a variety of traffic models. In addition, the thesis develops a method for traffic pattern classification in predictive channel selection. Classification-based prediction is shown to increase the throughput and reduce the number of collisions with the primary user up to 70% compared to the predictive system operating without classification. In terms of the power allocation work, the thesis defines the transmission power limit for secondary users as a function of the detection threshold of a spectrum sensor as well as investigates theoretical water-filling and truncated inverse power control methods. The methods have been optimized using rational decision theory concepts. The main focus has been on the development and performance comparison of practical inverse power control methods for constant data rate applications. One of the key achievements of the work is the development of the filtered-x LMS (FxLMS) algorithm based power control. It can be seen as a generalized inverse control to be used in power control research, giving a unified framework to several existing algorithms as well.
Original languageEnglish
QualificationDoctor Degree
Awarding Institution
  • University of Oulu
Supervisors/Advisors
  • Iinatti, Jari, Supervisor, External person
  • Mämmelä, Aarne, Supervisor
Award date12 Sep 2014
Publisher
Print ISBNs978-951-38-8267-9
Electronic ISBNs978-951-38-8268-6
Publication statusPublished - 2014
MoE publication typeG5 Doctoral dissertation (article)

Fingerprint

Frequency allocation
Radio systems
Cognitive radio
Power control
Frequency bands
Throughput
Decision theory
Power transmission
Fading channels
Maximum likelihood
Pattern recognition
Systems analysis
Sensors
Water

Keywords

  • dynamic spectrum access
  • prediction
  • closed-loop met

Cite this

@phdthesis{b6a6b6e2037b44bca97d514184bb3934,
title = "Adaptive power and frequency allocation strategies in cognitive radio systems: Dissertation",
abstract = "This doctoral thesis comprises a summary of novel results considering (1) channel selection in a cognitive radio system (CRS) using history information and (2) power allocation in a selected frequency band assuming a fading channel. Both can be seen as methods to manage interference between in-system users as well as to the users of other systems operating in the same geographical area and frequency band. Realization of CRSs that are using various methods to obtain information about environment and making intelligent decisions based on that information requires the use of adaptive transmission. Adaptive techniques proposed in this thesis enable efficient operation of CRSs in varying radio environment. History information and learning are essential factors to consider in the CRS design. Intelligent use of history information affects throughput, collisions and delays since it helps to guide the sensing and channel selection processes. In contrast to majority of approaches presented in the literature, this thesis proposes a classification-based prediction method that is not restricted to a certain type of traffic. Instead, it is a general method that is applicable to a variety of traffic classes. The work develops an optimal prediction rule for deterministic traffic pattern and maximum likelihood prediction rule for exponentially distributed traffic patterns for finding channels offering the longest idle periods for secondary operation. Series of simulations were conducted to show the general applicability of the rule to a variety of traffic models. In addition, the thesis develops a method for traffic pattern classification in predictive channel selection. Classification-based prediction is shown to increase the throughput and reduce the number of collisions with the primary user up to 70{\%} compared to the predictive system operating without classification. In terms of the power allocation work, the thesis defines the transmission power limit for secondary users as a function of the detection threshold of a spectrum sensor as well as investigates theoretical water-filling and truncated inverse power control methods. The methods have been optimized using rational decision theory concepts. The main focus has been on the development and performance comparison of practical inverse power control methods for constant data rate applications. One of the key achievements of the work is the development of the filtered-x LMS (FxLMS) algorithm based power control. It can be seen as a generalized inverse control to be used in power control research, giving a unified framework to several existing algorithms as well.",
keywords = "dynamic spectrum access, prediction, closed-loop met",
author = "Marko H{\"o}yhty{\"a}",
note = "Project code: 82120",
year = "2014",
language = "English",
isbn = "978-951-38-8267-9",
series = "VTT Science",
publisher = "VTT Technical Research Centre of Finland",
number = "61",
address = "Finland",
school = "University of Oulu",

}

Adaptive power and frequency allocation strategies in cognitive radio systems : Dissertation. / Höyhtyä, Marko.

VTT Technical Research Centre of Finland, 2014. 78 p.

Research output: ThesisDissertationCollection of Articles

TY - THES

T1 - Adaptive power and frequency allocation strategies in cognitive radio systems

T2 - Dissertation

AU - Höyhtyä, Marko

N1 - Project code: 82120

PY - 2014

Y1 - 2014

N2 - This doctoral thesis comprises a summary of novel results considering (1) channel selection in a cognitive radio system (CRS) using history information and (2) power allocation in a selected frequency band assuming a fading channel. Both can be seen as methods to manage interference between in-system users as well as to the users of other systems operating in the same geographical area and frequency band. Realization of CRSs that are using various methods to obtain information about environment and making intelligent decisions based on that information requires the use of adaptive transmission. Adaptive techniques proposed in this thesis enable efficient operation of CRSs in varying radio environment. History information and learning are essential factors to consider in the CRS design. Intelligent use of history information affects throughput, collisions and delays since it helps to guide the sensing and channel selection processes. In contrast to majority of approaches presented in the literature, this thesis proposes a classification-based prediction method that is not restricted to a certain type of traffic. Instead, it is a general method that is applicable to a variety of traffic classes. The work develops an optimal prediction rule for deterministic traffic pattern and maximum likelihood prediction rule for exponentially distributed traffic patterns for finding channels offering the longest idle periods for secondary operation. Series of simulations were conducted to show the general applicability of the rule to a variety of traffic models. In addition, the thesis develops a method for traffic pattern classification in predictive channel selection. Classification-based prediction is shown to increase the throughput and reduce the number of collisions with the primary user up to 70% compared to the predictive system operating without classification. In terms of the power allocation work, the thesis defines the transmission power limit for secondary users as a function of the detection threshold of a spectrum sensor as well as investigates theoretical water-filling and truncated inverse power control methods. The methods have been optimized using rational decision theory concepts. The main focus has been on the development and performance comparison of practical inverse power control methods for constant data rate applications. One of the key achievements of the work is the development of the filtered-x LMS (FxLMS) algorithm based power control. It can be seen as a generalized inverse control to be used in power control research, giving a unified framework to several existing algorithms as well.

AB - This doctoral thesis comprises a summary of novel results considering (1) channel selection in a cognitive radio system (CRS) using history information and (2) power allocation in a selected frequency band assuming a fading channel. Both can be seen as methods to manage interference between in-system users as well as to the users of other systems operating in the same geographical area and frequency band. Realization of CRSs that are using various methods to obtain information about environment and making intelligent decisions based on that information requires the use of adaptive transmission. Adaptive techniques proposed in this thesis enable efficient operation of CRSs in varying radio environment. History information and learning are essential factors to consider in the CRS design. Intelligent use of history information affects throughput, collisions and delays since it helps to guide the sensing and channel selection processes. In contrast to majority of approaches presented in the literature, this thesis proposes a classification-based prediction method that is not restricted to a certain type of traffic. Instead, it is a general method that is applicable to a variety of traffic classes. The work develops an optimal prediction rule for deterministic traffic pattern and maximum likelihood prediction rule for exponentially distributed traffic patterns for finding channels offering the longest idle periods for secondary operation. Series of simulations were conducted to show the general applicability of the rule to a variety of traffic models. In addition, the thesis develops a method for traffic pattern classification in predictive channel selection. Classification-based prediction is shown to increase the throughput and reduce the number of collisions with the primary user up to 70% compared to the predictive system operating without classification. In terms of the power allocation work, the thesis defines the transmission power limit for secondary users as a function of the detection threshold of a spectrum sensor as well as investigates theoretical water-filling and truncated inverse power control methods. The methods have been optimized using rational decision theory concepts. The main focus has been on the development and performance comparison of practical inverse power control methods for constant data rate applications. One of the key achievements of the work is the development of the filtered-x LMS (FxLMS) algorithm based power control. It can be seen as a generalized inverse control to be used in power control research, giving a unified framework to several existing algorithms as well.

KW - dynamic spectrum access

KW - prediction

KW - closed-loop met

M3 - Dissertation

SN - 978-951-38-8267-9

T3 - VTT Science

PB - VTT Technical Research Centre of Finland

ER -