Biological cluster identification for ultra-wideband multipath propagations

In this paper, we investigate the cluster identification of ultra-wideband (UWB) multipath propagations from a promising biological processing perspective. In the presented biological cluster extraction method, both the amplitude decay and time of arrival of UWB channel impulse response (CIR) are fully taken into considerations. Each resolvable multipath component is projected onto a two dimensional amplitude-time workspace, and then modeled as a virtual ant-agent. Thus, these ant-agents can move around in this 2-D space with a preference to the high local environment similarity. By establishing a subtle population similarity and specifying an efficient position adaptation strategy, cluster identification can be elegantly realized by the biological ant colony clustering (ACC) procedure. As the experimental simulations shown, the suggested algorithm can accurately and efficiently identify the involved multiple clusters in a completely automatic manner, which is of great importance to UWB channel modeling and parameters extractions.