A Research On Intelligent Decision Engine Of OFDM System

Intelligent Decision Engine(IDE) is the core module of Cognitive Radio(CR), through optimization, reasoning and learning function to achieve the best allocation of system resources; OFDM(Orthogonal Frequency Division Multiplexing) technology is a highly efficient multi-carrier modulation technique, which has been widely applied to the current mainstream wireless communication systems, and it is suitable for cognitive radio systems.From the application points of AI(artificial intelligence) techniques in OFDM system resource allocation, this paper studies the design of Intelligent Decision Engine. Based on the perception of external information, combining with traffic requirements and constraint rules, IDE uses optimization, learning and reasoning modules to make decisions intelligently to adaptively adjust and configure system resources. Research work is divided into the following four parts:The first part of this article mainly researches resource allocation in OFDM systems and AI algorithms. The optimization guidelines of OFDM system dynamic resource allocation are introduced, including Margin Adaptive(MA) and Rate Adaptive(RA); meanwhile, studies the principles and applications of eight kinds of AI algorithms respectively, including Genetic algorithm, Ant Colony algorithm, Simulated Annealing algorithm, Binary Particle Swarm, Hidden Markov model, Rule-based System, Case-based System and Artificial Neural Network, and comparative analysis of several optimization algorithms and of learning algorithms are given, which provides theoretical guidance for further research below.The second part of this article focuses on the design of the resource allocation algorithm with the goal of minimizing system transmit power. On the basis the principles and processing flow of single-user and multi-user OFDM system resource allocation algorithm based on water-filling greedy algorithm, combining with the analysis of the simulation results of the impact of adjustment of adaptive genetic operators and improved methods on performance of genetic search, OFDM systems resource allocation algorithm based on improved genetic algorithm is introduced. TDMA-OFDM, FDMA-OFDM, multi-user greedy water-filling algorithm, the basic genetic algorithm and improved genetic algorithms are used for resource allocation, and the statistical properties and mathematical complexity of five algorithms are compared.The third part of this paper focuses on the design of maximizing system capacity with joint allocation of subchannel and power. According to the study of two suboptimal subchannel and power allocation algorithm with two-step process, joint subchannel and power allocation algorithm based on genetic algorithm is proposed. The design of chromosome and genetic operations for two goals of maximizing system capacity and user capacity proportional fairness requirement and transmit power limit constraints are discussed in detail; the performance and mathematical complexity of two kinds of sub-optimal algorithms, genetic algorithms and modified genetic algorithm under a large number of statistical simulation are comparative analyzed.The fourth paper studies the design of IDE based on Genetic Algorithm and Case-based Reasoning. Joint design of optimization module and learning module is given, and simulation of several examples of scenarios to verify the function of the system are given: Based on the channel types and gains of outside environment which system perceives, according to the communication targets and constraints, through learning module and decision module, IDE adaptively adjust subcarriers, bits, transmit power parameters, scheduling cycle and other parameters.