Improved Hysteretic Noisy Chaotic Neural Network for Broadcast Scheduling Problem in WMNs

Abstract

It has been proven that the noise-tuning-based hysteretic noisy chaotic neural network (NHNCNN) can use the noise tuning factor to improve the optimization performance obviously at lower initial noise levels while can not at initial higher noise levels. In order to improve the optimization performance of the NHNCNN at initial higher noise levels, we introduce a new noise tuning factor into the activation function and propose an improved hysteretic noisy chaotic neural network (IHNCNN) model. By regulating the value of the newly introduced noise tuning factor, both noise levels of the activation function and hysteretic dynamics in the IHNCNN can be adjusted to help to improve the global optimization ability as the initial noise amplitude is higher. As a result, the IHNCNN can exhibit better optimization performance at initial higher noise levels. In order to demonstrate the advantage of the IHNCNN over the NHNCNN, the IHNCNN combined with gradual expansion scheme (GES) is applied to solve broadcast scheduling problem (BSP) in wireless multihop networks (WMNs). The aim of BSP is to design an optimal time-division multiple-access (TDMA) frame structure with minimal frame length and maximal channel utilization. Simulation results in BSP show the superiority of the IHNCNN.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i3.2231