| Thte future mobile communication refers to the next generation mobile communication system. Dynamic resource assignment aims to more effectively make use of the existing spectrum resource anmong cells considering non-uniform traffic loads and fading channel condition, which is crucial to the future mobile communications.Dynamic raido resource assignment refers to the dynamic assigning available spectrum resource to users on condition that the traffic loads among cells are not uniform and the channels are fading inside cells, which is able to effectively improve the spectrum utilization.This doctoral thesis focuses on solving dynamic radio resource assignment problem in the future mobile communications, including dynamic channel (or frequency) assignment among cells, and subcarrier, bit, power assignment for a multiuser OFDM system inside a cell, which are the typical Nondeterministic Polynomial complete (NP-Complete) problem, whose computational complexity will increase exponentially with the number of object increasing. The existing algorithm for sovling such an NP-Complete ploblem includes Stochastic Simulated Annealing (SSA), Genetic Algorithm (GA), Combinatorial Evolution Strategy (CES), Tabu Search (TS), and Artificial Neural Network (ANN), etc. ANN can be implemented by hardware, and its computational time is up to a nano-second, which can meet the real-time demand of the mobile communications. Forthemore, the computational time of ANN will not increase with the number of object increasing.In recent year, it is report that the Noisy Chaotic Neural Network (NCNN) performs much better that the other type of ANN, whose search probability for an optimal solution of10-city Traveling Salesman Problem (TSP) is up to99.4%. Therefore, this thesis employs the NCNN to solve the dynamic radio resource assignment problem for the future mobile communications.The main contributions of this thesis are as follows:1) This thesis presents a novel Dynamic Channel Assignment (DCA) algorithm using NCNN for solving the small-scale cellular networks. In this algorithm, some deficiencies of the existing energy functions are considered, and then a novel energy function of DCA is proposed, which not only considers the three interference constraints (Co-Site Constraint, Adjacent Channel Constraint, Co-Channel Constraint) among cells and the required number of channels in each cell, but also minimizes the total number of assigned channels to improve the spectrum utilization. In addition, the novel energy function avoids generating an inflexible compatibility matrix in contrast to the existing energy function. So each punishment parameter can be adjusted flexibly, which makes the NCNN converges faster.2) This thesis presents a novel DCA algorithm for the large-scale cellular networks using NCNN.To our best knowledge, the existing DCA algorithm only focuses on the small-scale cellular networks,in which the total number of cells is no more than49.However, the practical cellular networks are all in a larger scale with hundreds or even thousands of cells. This thesis first divides a large-scale cellular network into many distributed cellular subnets, and then presents an improved energy function. Finally, an interference channel table is formulated for each subnet, which contains interference information of its neighboring subnets.The DCA process is simply implemented in each subnet independently according to its interference channel table, which avoids the mutual interference among subnets, and avoids the mutual interference among cells inside a subnet as well.As a result, the performance of the DCA for a large-scale cellular network is maintained on a level of that in small-scale cellular networks.3) This thesis presents a novel Dynamic Spectrum Allocation (DSA) algorithm for Cognitive Radio Cellular Networks (CRCNs) using NCNN. To our best knowledge, the existing DSA algorithm for cognitive radio only focuses on the Ad hoc networks, Mesh networks, etc. The traditional DSA schemes of general cellular networks are not suitable for CRCNs, in which the available frequencies (unoccupied by primary users) are not fixed and constantly change over time from one region to another. This thesis presents a novel DSA algorithm by employing the energy function which is proposed in the DCA algorithm for large-scale cellular networks. In the proposed DSA algorithm, an interference frequency table is formulated for CRCNs, which contains the real-time occupied frequencies of primary users. As result, the harmful interference to primary users is avoided, and the three mutual interference constraints among cells and the number of required frequencies of each cell are satisfied, and the total number of assigned frequencies from the available frequencies is minimized as well. In addition, the proposed DSA algorithm relatively fixedly assigns a frequency to a cell as the dedicated common control channel by selecting a frequency with the lowest serial number from the assigned frequencies, which controls all the cognitive users in a cell.4) This thesis presents a novel dynamic radio resource (including subcarrier, bit, and power) assignment algorithm for multiuser OFDM systems. To solve this problem, the existing schemes include two main types, which are the Margin Adaptive (MA) optimization and the Rate Adaptive (RA) optimization. The former refers to minimizing the total consumed power under a data rate constraint, whereas the latter refers to maximizing the data rate under a power constraint. It is reported that the RA opitimization can be replaced by recursive MA optimization. Therefore, this thesis only focuses on MA optimization problem. The existing MA optimization schemes are all with limited efficience. This thesis presents a novel energy function of MA optimization for multiuser OFDM systems using NCNN, which imporves the performance of MA optimization comparied with the existing schemes, and meets the real-time demand of the future mobile communications. |
PDF Full Text Request