A Modified Whale Optimization Algorithm And Its Application In Cognitive Radio Spectrum Allocation

Compared with traditional deterministic optimization algorithm,meta-heuristic algorithm is especially suitable for high dimensional optimization problems because of the characteristics of wide application range,no need for gradient information,strong global search ability and strong parallelism.Thus,meta heuristic algorithm has more and more applications in practical engineering problems.Whale optimization algorithm(WOA)belongs to the category of meta-heuristic algorithm,and it has received considerable attention from practitioners and researchers due to its easy implement and high precision.But it still has demerits with regard to slow convergence and easily trapped into local optimum.In this thesis,WOA is analyzed and researched deeply,and a modified whale optimization algorithm(MWOA)is proposed to overcome these shortcomings.The following improvements have been made:(1)The encircling threshold in WOA is fixed,thus in the later stage of iteration,the position update strategy is single and lack of flexibility.Self-adapting encircling threshold is introduced into the control parameter,individuals can adjust their encircling threshold according to their fitness values,which making the position update mode more flexible.(2)Other whales are directed only by current optimal solution in the spiral mechanism,and there is no mechanism to jump out of local optima.Consequently,WOA is easy to be trapped in premature phenomena.The learning mechanism of quantum particle swarm optimization(QPSO)is introduced to MWOA.In this phase,spiral mechanism is abandoned,current optimal whale and personal best position guide the position update and quantum-behavior whale is introduced simultaneously,which could expand the exchange of information and enhance the diversity of the population.(3)In addition,mean best position is utilized to guide other whales finding the optimal solution.The mean best position reflects the average value of the best positions that whales experienced in search process,which helps the whales move to its vicinity.This improvement can accelerate the convergence speed.Numerical optimization experiments are conducted on a set of benchmark functions.The simulation results verify the effectiveness and the advancement of MWOA in terms of convergence rate,precision and robustness.Furthermore,this thesis applies WOA in spectrum allocation problem of cognitive radio based on graph theory.but the results indicate that the binary whale algorithm(BWOA)has poor performance such as long computation time,slow convergence speed,low convergence precision.To tackle these issues,a genetic whale algorithm(GWOA)is proposed in this thesis.Some improvements have been done as follows:(1)Remove the discretization in BWOA,and cut off the mapping from continuous domain to discrete domain.This tactic can reduce computation load.(2)Abandon the continuous domain position update formula in WOA,crossover and mutation operation in genetic algorithm(GA)is adopted in the stage of exploration and can shorten the searching time,which is directly executed in the discrete domain.In the stage of exploitation,one-dimensional updating is used to guarantee the otherness of new solution and strengthen the local searching ability of the algorithm.(3)Well balancing the exploitation and exploration ability of the algorithm,a self-adapt selection threshold is introduced in GWOA,which make the improved algorithm has better global search ability in the early stage,and has stronger local searching ability in the later stage.The GWOA is applied to the spectrum allocation problem,and simulation experiments are conducted with other intelligent algorithms.The simulation results show that the proposed algorithm has higher network reward and faster convergence speed,and is more suitable for the problem of spectrum allocation when compared with other intelligent algorithms.