The Research And Application Of Adaptive Quantum Bacterial Foraging Optimization

Swarm intelligence algorithm is a hot research topic in the field of artificial intelligence in recent years. With the advantages of simple structure, flexible, robust and strong ability of self-organization, the bacterial foraging optimization(BFO) algorithm has been widely concerned in the international computational intelligence community. However, the convergence of the BFO algorithm is slow and easy to fall into local optimum for complex multimodal functions. Aiming at this problem, this paper improves the BFO algorithm through the perspective of more realistically mimics the bacterial behavior and introducing the quantum computing respectively. And the improved algorithm has been used in the design of robust training sequences in MIMO-OFDM channel of the 4G mobile communication. The main contents of this paper are as follows:First, the depth study of the classical BFO algorithm is carried out, and it is found that the progress of the bacteria is forward to the random direction along the straight line. However, this is not consistent with the true bacterial foraging process. To solve this problem, this paper makes the position of the bacteria to the current population of the optimal bacteria. The typical benchmark functions are tested and the experimental results show that the improved algorithm can speed up the search speed and improve the local search ability.Second, the performance of the quantum bacterial foraging optimization(QBFO) is studied, and the results show that the optimization results are better than the classical BFO algorithm. A comparative analysis of the main parameters of the rotating angle is carried out and the results show that the different rotation angles of the final optimization results are very different from the experimental results; the fixed rotational phase is one of the main factors that affect the performance of the QBFO algorithm.Third, a new method for the design of the adaptive rotation phase of the QBFO algorithm, so that the angle of the rotating phase of the current and the current bacteria and the angle difference. Through 16 different types of standard test functions of optimization performance research, statistical results show that the algorithm in the low dimensional, convergence accuracy and stability is superior to prior to the improvement of algorithm, and the optimization result is obviously better than the classical BFO and quantum genetic algorithm(QGA). Further research shows that, average convergence probability of the proposed algorithm is the highest and the average running time and average running steps are the shortest among the four algorithms when reach the specified convergence precision.Finally, the problem of designing robust training sequence in MIMO channel estimation is analyzed and the problem is a kind of min-max problem essentially. In view of the advantages of QBFO in complex nonconvex optimization, a method based on QBFO algorithm for solving the robust sequence is proposed. The simulation results show that this new algorithm and the existing robust design based on iterative algorithm, in both cases of Kronecker channel and any relevant channel, the former algorithm is smaller to the latter in the mean square of the estimated channel, especially in the situation of low SNR.