A Research Of Radio Resource Allocation Based On Fireworks Algorithm In Green LTE Cellular Systems

With the increasing requirements of high-rate services as well as the rapidly-expanded scale of networks, the energy consumption of wireless communication systems skyrockets. With the arrival of 4G mobile communication, the situation has turned serious. The statistics show that two-thirds of the voice calls, as well as over 90% of data traffic occurred indoors. To offset the serious loss of wireless signal, however, we have got to increase the transmission power of base station so as to guarantee the service quality of the users indoors, which exacerbate the energy consumption in macro cellular networks.Radio resource management for high energy efficiency is an effective way to alleviate energy consumption in wireless communication systems. Above this, the thesis is to introduce Femtocell, with which we can establish a green LTE cellular networks, to conduct a meaningful research on radio resource optimization for green communication.Making use of game theory, the thesis is to construct a utility function reflecting energy efficiency based on sigmoid function whose value diminishes marginally. By optimizing the allocation of power and resource block, base stations will maximize the energy efficiency of downlink channel. However, since the joint optimization of power and resource block in green cellular networks is an NP-hard problem, it is really complex to solve. By relaxing the constraint conditions, we can prove the optimization problem is a quasi concave problem which has globally optimal solutions. We optimize the allocation of resource block and power in turn with iterations and introduce swarm algorithms to accelerate the convergence of the optimization.The thesis is going to recommend a new swarm algorithm called fireworks algorithm, which is an excellent algorithm. By mathematical derivation, we conclude the global convergence of it. In other words, given enough time, it will always get the global optimums, regardless the trap of local optimums. By numerical analysis towards five test functions, we get the conclusion that fireworks algorithm outperforms particle swarm algorithm and genetic algorithm.At last, the system-level simulator of University of Vienna based on FDD-LTE is chosen to simulate the convergence and effectiveness of the proposed optimization problem. By adding scheduler functions based on different swarm algorithms, it is easy to conclude that the proposed resource optimization converges and outperforms the classic scheduling algorithms, namely, round robin algorithm, max C/I algorithm and proportional fair algorithm.