Title Page
Abstract
Contents
Chapter 1. Introduction 19
1.1. Radio Environments Overview 20
1.2. Cognitive Radio (CR): A Future Radio System 20
1.3. Classification of Resource Allocation in CR 22
1.4. Motivations and Contributions 35
1.5. Dissertation Organization 37
Chapter 2. Graph-theory Based Cooperative Spectrum Allocation in CR Networks Using Bipartite Matching 38
2.1. Background 39
2.2. Problem Formulation and Network Modeling 41
2.3. Bipartite Matching Algorithms 44
2.4. Numerical Results and Discussions 57
2.5. Summary of The Chapter 62
Chapter 3. Power Allocation in OFDM-Based CR Networks 65
3.1. Background 66
3.2. System Model and SU’s QoS Assurance 68
3.3. Power Allocation in Single-SU Case 71
3.4. Power Allocation in Multiple-SUs Case 98
3.5. Summary of The Chapter 103
Chapter 4. An Admission Control Scheme in CR Networks 109
4.1. Background 110
4.2. System Model and Denotations 112
4.3. PU’s Receiver Detectable Case 116
4.4. PU’s Receiver Undetectable Case 118
4.5. Admission Control Algorithm for SUs 122
4.6. Numerical Results and Discussions 123
4.7. Summary of The Chapter 129
Chapter 5. Joint Sensing Time and Power Allocation in CR Networks with Amplify-and-Forward Cooperation 130
5.1. Background 131
5.2. System Model and Formulation 133
5.3. Optimization without Power Limitation 136
5.4. Optimization with Power Limitation 139
5.5. Summary of The Chapter 146
Chapter 6. Conclusions and Future Work 148
6.1. Conclusions 148
6.2. Future Work 149
References 151
Publications 158
Table 2.1: Complexities of the algorithms. 57
Table 3.1: Simulation parameters 93
Figure 1.1: A snapshot of the spectrum utilization in an urban area. 20
Figure 1.2: Cognitive computer cycle. 21
Figure 1.3: The hierarchy for the resource management in CR systems. 22
Figure 1.4: A demonstration of spectrum underlay sharing. 25
Figure 1.5: Opportunistic sharing concept of the spectrum. 26
Figure 1.6: An example of sensing slot scheduling in a CR network. 34
Figure 2.1: An example of the edge-vertex transform. 42
Figure 2.2: An example of perfect matching. 45
Figure 2.3: Sharing-considered bipartite graph. 47
Figure 2.4: Repetition-deleted bipartite graph. 48
Figure 2.5: Matching algorithm by externally improving on K-M algorithm, where step 1 is to resolve the starving problem and step 2 is to resolve the sharing problem. 50
Figure 2.6: Starving user matched bipartite graph. 51
Figure 2.7: Maximum match of labeled spanning subgraph. 52
Figure 2.8: Matches happen in the 2nd round. 53
Figure 2.9: Matching algorithm by internally improving on K-M algorithm, where the function of Ngl(X) is to obtain the neighboring vertices of set X in a labeled spanning subgraph. 55
Figure 2.10: Functions. 56
Figure 2.11: The influence of α on the performances, where N = 20, M = 10 and β = 0.5; utility per user = network utility/number of users; user starvation probability = amount of starved users / number of users.(이미지참조) 60
Figure 2.12: The influence of β on the performances, where N = 20, M = 10 and α = 0.5. 61
Figure 3.1: PU protection schemes. 69
Figure 3.2: Considered cognitive network model. 72
Figure 3.3: The maximum transmission power in the SU’s subcarriers constrained by the CCI. 75
Figure 3.4: Convergence of the algorithm corresponding to ω*.(이미지참조) 84
Figure 3.5: Peak power limited single user water-filling, the area under the line of [CGNR-1 n + Emax n , E0]- and beyond the line of [CGNR-1 n , E0]- represents the total allocated power.(이미지참조) 85
Figure 3.6: Graphical illustration of the 2nd step of the proposed algorithm, where f (n) = CGNR-1 n and g(n) = CGNR-1 n + Emax(이미지참조) 87
Figure 3.7: Graphical illustration of the 3rd step of the proposed algorithm, where f (n) = CGNR-1 n and g(n) = CGNR-1 n + Emax(이미지참조) 89
Figure 3.8: Simulation scene. 93
Figure 3.9: Subcarrier occupations. 94
Figure 3.10: Maximum transmission power constrained by CCI and ACI, where only the subcarriers 1∼16 are depicted. 94
Figure 3.11: Noise plus ACI interference induced by PUs on subcarriers 1∼16. 95
Figure 3.12: The concept of “cap-limited” water-filling. 96
Figure 3.13: Bit-loading without integral-bit constraint. 96
Figure 3.14: SU’s capacity on subcarriers used by different PUs. Where, “SCs” stands for “subcarriers”, Ebudget = 145dB × 1024 in “Case 1” and Ebudget = 135dB × 1024 in “Case 2(이미지참조) 97
Figure 3.15: Water-filling in multiple-SUs case. Subcarrier n is allocated to user arg maxs [[1 - CGNR-1 s,n, 0]+, Emax s,n - , e.g., subcarrier n is allocated to user 1.(이미지참조) 99
Figure 3.16: Simulation scenario. 101
Figure 3.17: Gs,n for 4 secondary users according to COST-207 model, where the results are shown in dB compared with σ2.(이미지참조) 102
Figure 3.18: Power allocation for 4 secondary users, where thick lines denote CGNR-1 s,n, and thin lines denote CGNR-1 s,n + Emax s,n . (A logarithmic y-axis is shown in the figure, where “0dB” corresponds to the “1” in Fig. 3.15.)(이미지참조) 103
Figure 4.1: Considered cognitive network in PU’s receiver detectable case. 113
Figure 4.2: Spectrum opportunity detection: A conservative approach that transforms detecting primary receiver to detecting primary transmitter. 119
Figure 4.3: Considered cognitive network in primary receiver undetectable case. 121
Figure 4.4: Upper bound of PmGm,PU/σ2 and lower bound of PmGm,m/σ2.(이미지참조) 124
Figure 4.5: Upper bound of Gm,PU/Gm,m.(이미지참조) 125
Figure 4.6: Restriction lines on the distance between primary receiver and secondary transmitter 126
Figure 4.7: Restriction lines on the distance between primary and secondary transmitters 127
Figure 5.1: Model for cooperative sensing and transmission. 133
Figure 5.2: Frame structure and power allocation of the considered cooperation. 133
Figure 5.3: Achievable throughput for the cooperative secondary network. 139
Figure 5.4: Optimal values of β₁ and β₂. 144
Figure 5.5: Achievable throughput for the cooperative secondary network, where β₁ = 1 and β₂ = 0.523. 145
Figure 5.6: Optimal values of β₁ and β₂. 146
Figure 5.7: Achievable throughput for the cooperative secondary network, where β₁ = 0.946 and β₂ = 0.550. 147