基于非线性因子的改进鸟群算法在动态能耗管理中的应用

Application of Improved Bird Swarm Algorithm Based on Nonlinear Factor in Dynamic Energy Management

针对实时系统能耗管理中动态电压调节(DVS)技术的应用会导致系统可靠性下降的问题,该文提出一种基于改进鸟群(IoBSA)算法的动态能耗管理法。首先,采用佳点集原理均匀地初始化种群,从而提高初始解的质量,有效增强种群多样性;其次,为了更好地平衡BSA算法的全局和局部搜索能力,提出非线性动态调整因子;接着,针对嵌入式实时系统中处理器频率可以动态调整的特点,建立具有时间和可靠性约束的功耗模型;最后,在保证实时性和稳定性的前提下,利用提出的IoBSA算法,寻求最小能耗的解决方案。通过实验结果表明,与传统BSA等常见算法相比,改进鸟群算法在求解最小能耗上有着很强的优势及较快的处理速度。

The application of Dynamic Voltage Scaling(DVS) technique in real-time system energy management will result in the decrease of system reliability. A dynamic energy management method based on Improved Bird Swarm Algorithm(IoBSA) is proposed in this paper. Firstly, the population is initialized uniformly with the principle of good point set, so as to improve the quality of initial solution and increase the diversity of population effectively. Secondly, in order to balance better the global and local search ability of BSA algorithm, the nonlinear dynamic adjustment factor is proposed. Then, a power consumption model with time and reliability constraints is established for the dynamic adjustment of processor frequency in embedded real-time systems. On the premise of ensuring real-time performance and stability, the proposed IoBSA algorithm is used to find the solution with minimum energy consumption. The experimental results show that compared with the traditional BSA algorithm and other common algorithms, the improved bird swarm algorithm has a strong advantage in solving the minimum energy consumption and a fast processing speed energy management.